STEINGEN/docs/Articles.bib

@article{aboukasmIntegratedBerthAllocation2020,
  title = {The Integrated Berth Allocation, Quay Crane Assignment and Scheduling Problem: Mathematical Formulations and a Case Study},
  shorttitle = {The Integrated Berth Allocation, Quay Crane Assignment and Scheduling Problem},
  author = {Abou Kasm, Omar and Diabat, Ali and Cheng, T. C. E.},
  date = {2020-08-01},
  journaltitle = {Annals of Operations Research},
  shortjournal = {Ann Oper Res},
  volume = {291},
  number = {1},
  pages = {435--461},
  issn = {1572-9338},
  doi = {10.1007/s10479-018-3125-3},
  url = {https://doi.org/10.1007/s10479-018-3125-3},
  abstract = {This paper considers the integration of three essential seaport terminal operations: the berth allocation problem, the quay crane assignment problem (QCAP), and the quay crane scheduling problem (QCSP). The paper presents a new mathematical formulation that captures all associated operations and constraints. Different quay crane operational policies are considered, namely permitting versus not permitting bay task preemption in QCSP and static versus dynamic crane allocations in QCAP. Thus, variants of the mathematical formulation are introduced to capture the different combinations of these scenarios. Due to the preemption consideration, the models include disaggregated quay crane (QC) tasks. Specifically, QC tasks are identified by single container movements as opposed to bay or stack task allocations that are commonly used in the literature. A case study based on Abu Dhabi’s container terminal is presented where the use of the proposed mathematical models are compared against the current existing operational approach. Results show that the service times can be significantly decreased by the use of the proposed models. Moreover, the policy choice effect on the total schedule is compared through simulated examples and Abu Dhabi’s container terminal case study. The results show that the policy improvements can depend on the problem’s attributes and thus a better policy cannot be generalized.},
  langid = {english},
  keywords = {Berth allocation problem,Maritime logistics,Optimization modeling,Quay crane assignment,Quay crane scheduling,Seaport operations},
  file = {C:\Users\sjelic\Zotero\storage\IC8BAKSS\Abou Kasm et al. - 2020 - The integrated berth allocation, quay crane assign.pdf}
}

@article{ackermanWeightedClustering2021,
  title = {Weighted {{Clustering}}},
  author = {Ackerman, Margareta and Ben-David, Shai and Brânzei, Simina and Loker, David},
  date = {2021-09-20},
  journaltitle = {Proceedings of the AAAI Conference on Artificial Intelligence},
  shortjournal = {AAAI},
  volume = {26},
  number = {1},
  pages = {858--863},
  issn = {2374-3468, 2159-5399},
  doi = {10.1609/aaai.v26i1.8282},
  url = {https://ojs.aaai.org/index.php/AAAI/article/view/8282},
  abstract = {In this paper we investigate clustering in the weighted setting, in which every data point is assigned a real valued weight. We conduct a theoretical analysis on the influence of weighted data on standard clustering algorithms in each of the partitional and hierarchical settings, characterising the precise conditions under which such algorithms react to weights, and classifying clustering methods into three broad categories: weight-responsive, weight-considering, and weight-robust. Our analysis raises several interesting questions and can be directly mapped to the classical unweighted setting.},
  langid = {english},
  file = {C:\Users\sjelic\Zotero\storage\WHYVNW33\Ackerman et al. - 2021 - Weighted Clustering.pdf}
}

@inproceedings{akhmedovFastPrizeCollectingSteiner2017,
  title = {A {{Fast Prize-Collecting Steiner Forest Algorithm}} for {{Functional Analyses}} in {{Biological Networks}}},
  booktitle = {Integration of {{AI}} and {{OR Techniques}} in {{Constraint Programming}}},
  author = {Akhmedov, Murodzhon and LeNail, Alexander and Bertoni, Francesco and Kwee, Ivo and Fraenkel, Ernest and Montemanni, Roberto},
  editor = {Salvagnin, Domenico and Lombardi, Michele},
  date = {2017},
  series = {Lecture {{Notes}} in {{Computer Science}}},
  pages = {263--276},
  publisher = {Springer International Publishing},
  location = {Cham},
  doi = {10.1007/978-3-319-59776-8_22},
  abstract = {The Prize-collecting Steiner Forest (PCSF) problem is NP-hard, requiring extreme computational effort to find exact solutions for large inputs. We introduce a new heuristic algorithm for PCSF which preserves the quality of solutions obtained by previous heuristic approaches while reducing the runtime by a factor of 10 for larger graphs. By decreasing the draw on computational resources, this algorithm affords systems biologists the opportunity to analyze larger biological networks faster and narrow their analyses to individual patients.},
  isbn = {978-3-319-59776-8},
  langid = {english},
  keywords = {Biological networks,Prize-collecting Steiner Forest}
}

@article{alfandariTailoredBendersDecomposition2022,
  title = {A Tailored {{Benders}} Decomposition Approach for Last-Mile Delivery with Autonomous Robots},
  author = {Alfandari, Laurent and Ljubić, Ivana and De Melo da Silva, Marcos},
  date = {2022-06-01},
  journaltitle = {European Journal of Operational Research},
  shortjournal = {European Journal of Operational Research},
  volume = {299},
  number = {2},
  pages = {510--525},
  issn = {0377-2217},
  doi = {10.1016/j.ejor.2021.06.048},
  url = {https://www.sciencedirect.com/science/article/pii/S0377221721005646},
  abstract = {This work addresses an operational problem of a logistics service provider that consists of finding an optimal route for a vehicle carrying customer parcels from a central depot to selected facilities, from where autonomous devices like robots are launched to perform last-mile deliveries. The objective is to minimize a tardiness indicator based on the customer delivery deadlines. This article provides a better understanding of how three major tardiness indicators can be used to improve the quality of service by minimizing the maximum tardiness, the total tardiness, or the number of late deliveries. We study the problem complexity, devise a unifying Mixed Integer Programming formulation and propose an efficient branch-and-Benders-cut scheme to deal with instances of realistic size. Numerical results show that this novel Benders approach with a tailored combinatorial algorithm for generating Benders cuts largely outperforms all other alternatives. In our managerial study, we vary the number of available facilities, the coverage radius of autonomous robots and their speed, to assess their impact on the quality of service and environmental costs.},
  langid = {english},
  keywords = {Benders decomposition,Integer Programming,Last-mile delivery,Self-driving robots},
  file = {C\:\\Users\\sjelic\\Zotero\\storage\\NR9Q65RV\\Alfandari et al. - 2022 - A tailored Benders decomposition approach for last.pdf;C\:\\Users\\sjelic\\Zotero\\storage\\RUB9TPIK\\S0377221721005646.html}
}

@thesis{alsoufiMathematicalModelsSeaside2017,
  type = {phdthesis},
  title = {Mathematical {{Models}} of {{Seaside Operations}} in {{Container Ports}} and Their {{Solution}}},
  author = {Alsoufi, Ghazwan},
  date = {2017-07},
  institution = {University of Essex},
  url = {https://repository.essex.ac.uk/20013/},
  abstract = {Operational Research and Optimization are fundamental disciplines which, for decades, provided the real-world with tools for solving practical problems. Many such problems arise in container ports. Container terminals are important assets in modern economies. They constitute an important means of distributing goods made overseas to domestic markets in most countries. They are expensive to build and difficult to operate. We describe here some of the main operations which are faced daily by decision makers at those facilities. Decision makers often use Operational Research and Optimization tools to run these operations effectively. In this thesis, we focus on seaside operations which can be divided into three main problems: 1- the Berth Allocation Problem (BAP), 2- the Quay Crane Assignment Problem (QCAP), 3- the Quay Crane Scheduling Problem (QCSP). Each one of the above is a complex optimization problem in its own right. However, solving them individually without the consideration of the others may lead to overall suboptimal solutions. For this reason we will investigate the pairwise combinations of these problems and their total integration In addition, several important factors that affected on the final solution. The main contributions of this study are modelling and solving of the: 1- Robust berth allocation problem (RBAP): a new efficient mathematical model is formulated and a hybrid algorithm based on Branch-and-Cut and the Genetic Algorithm is used to find optimal or near optimal solutions for large scale instances in reasonable time. 2- Quay crane assignment and quay crane scheduling problem (QCASP): a new mathematical model is built to simultaneously solve QCASP and a heuristic based on the Genetic Algorithm is developed to find solutions to realistic instances in reasonable time. 3- Berth allocation, quay crane assignment and quay crane scheduling problem (BACASP): an aggregate model for all three seaside operations is proposed and to solve realistic instances of the problem, an adapted variant of the Genetic Algorithm is implemented. Keywords: berth allocation; quay crane assignment; quay crane scheduling; terminal operations; genetic algorithm},
  langid = {english},
  pagetotal = {138},
  file = {C\:\\Users\\sjelic\\Zotero\\storage\\J6T3QZ3K\\Alsoufi - 2017 - Mathematical Models of Seaside Operations in Conta.pdf;C\:\\Users\\sjelic\\Zotero\\storage\\8DEJWTEA\\20013.html}
}

@article{anDCDifferenceConvex2005,
  title = {The {{DC}} ({{Difference}} of {{Convex Functions}}) {{Programming}} and {{DCA Revisited}} with {{DC Models}} of {{Real World Nonconvex Optimization Problems}}},
  author = {An, Le Thi Hoai and Tao, Pham Dinh},
  date = {2005-01-01},
  journaltitle = {Annals of Operations Research},
  shortjournal = {Ann Oper Res},
  volume = {133},
  number = {1},
  pages = {23--46},
  issn = {1572-9338},
  doi = {10.1007/s10479-004-5022-1},
  url = {https://doi.org/10.1007/s10479-004-5022-1},
  abstract = {The DC programming and its DC algorithm (DCA) address the problem of minimizing a function f=g−h (with g,h being lower semicontinuous proper convex functions on Rn) on the whole space. Based on local optimality conditions and DC duality, DCA was successfully applied to a lot of different and various nondifferentiable nonconvex optimization problems to which it quite often gave global solutions and proved to be more robust and more efficient than related standard methods, especially in the large scale setting. The computational efficiency of DCA suggests to us a deeper and more complete study on DC programming, using the special class of DC programs (when either g or h is polyhedral convex) called polyhedral DC programs. The DC duality is investigated in an easier way, which is more convenient to the study of optimality conditions. New practical results on local optimality are presented. We emphasize regularization techniques in DC programming in order to construct suitable equivalent DC programs to nondifferentiable nonconvex optimization problems and new significant questions which have to be answered. A deeper insight into DCA is introduced which really sheds new light on DCA and could partly explain its efficiency. Finally DC models of real world nonconvex optimization are reported.},
  langid = {english},
  keywords = {DC algorithms (DCA),DC duality,DC programming,global optimality conditions,local optimality conditions,polyhedral DC programming,regularization techniques},
  file = {C:\Users\sjelic\Zotero\storage\22IW9JP5\An и Tao - 2005 - The DC (Difference of Convex Functions) Programmin.pdf}
}

@inproceedings{angunNewMixedIntegerLinear2019,
  title = {A {{New Mixed-Integer Linear Programming Formulation}} for {{Multiple Responses Regression Clustering}}},
  booktitle = {2019 6th {{International Conference}} on {{Control}}, {{Decision}} and {{Information Technologies}} ({{CoDIT}})},
  author = {Angün, Ebru and Altınoy, Alper},
  date = {2019-04},
  pages = {1634--1639},
  issn = {2576-3555},
  doi = {10.1109/CoDIT.2019.8820674},
  abstract = {This paper considers a regression-based clustering problem for multiple responses. We propose a nested optimization formulation, where two minimizations are performed one after the other. The first minimization is to determine the number of clusters, and the second is to minimize the L∞-norm of residuals. After fixing the number of clusters, our formulation reduces to a novel Mixed-Integer Linear Programming (MILP) problem, which can handle multiple responses simultaneously. We further propose an empirical approach based on cross-validation to determine a good number of clusters; this approach takes into account prediction accuracies of models when choosing the number of clusters. Using the JURA dataset, we illustrate that the classic approach in the literature, which considers one response at-a-time, usually assigns the same entity to different clusters with respect to different responses; hence, eventually, it is not evident to which cluster that entity belongs to. Also, even though the classic approach assigns that entity to the same cluster with respect to all responses, this assignment can be different than the one obtained when all responses are considered simultaneously; hence, the classic approach can result in a false clustering when multiple responses have to be considered at the same time.},
  eventtitle = {2019 6th {{International Conference}} on {{Control}}, {{Decision}} and {{Information Technologies}} ({{CoDIT}})},
  keywords = {Approximation algorithms,Heuristic algorithms,Linear programming,Linear regression,Minimization,Predictive models,Upper bound},
  file = {C\:\\Users\\sjelic\\Zotero\\storage\\CIIF5PQ2\\Angün и Altınoy - 2019 - A New Mixed-Integer Linear Programming Formulation.pdf;C\:\\Users\\sjelic\\Zotero\\storage\\AZPKEWBV\\8820674.html}
}

@article{antonijevicTransferLearningApproach2023,
  title = {Transfer Learning Approach Based on Satellite Image Time Series for the Crop Classification Problem},
  author = {Antonijević, Ognjen and Jelić, Slobodan and Bajat, Branislav and Kilibarda, Milan},
  date = {2023-04-29},
  journaltitle = {Journal of Big Data},
  shortjournal = {Journal of Big Data},
  volume = {10},
  number = {1},
  pages = {54},
  issn = {2196-1115},
  doi = {10.1186/s40537-023-00735-2},
  url = {https://doi.org/10.1186/s40537-023-00735-2},
  abstract = {This paper presents a transfer learning approach to the crop classification problem based on time series of images from the Sentinel-2 dataset labeled for two regions: Brittany (France) and Vojvodina (Serbia). During preprocessing, cloudy images are removed from the input data, the time series are interpolated over the time dimension, and additional remote sensing indices are calculated. We chose TransformerEncoder as the base model for knowledge transfer from source to target domain with French and Serbian data, respectively. Even more, the accuracy of the base model with the preprocessing step is improved by 2\% when trained and evaluated on the French dataset. The transfer learning approach with fine-tuning of the pre-trained weights on the French dataset outperformed all other methods in terms of overall accuracy 0.94 and mean class recall 0.907 on the Serbian dataset. Our partially fine-tuned model improved recall of crop types that were poorly classified by the base model. In the case of sugar beet, class recall is improved by 85.71\%.},
  keywords = {Attention mechanism,Crop classification,Domain adaptation,Encoder–decoder architecture,Remote sensing,Transfer learning},
  file = {C\:\\Users\\sjelic\\Zotero\\storage\\FNAXHNZE\\Antonijević et al. - 2023 - Transfer learning approach based on satellite imag.pdf;C\:\\Users\\sjelic\\Zotero\\storage\\KYW7DBHZ\\s40537-023-00735-2.html}
}

@article{applebyKrigingConvolutionalNetworks2020,
  title = {Kriging {{Convolutional Networks}}},
  author = {Appleby, Gabriel and Liu, Linfeng and Liu, Li-Ping},
  date = {2020-04-03},
  journaltitle = {Proceedings of the AAAI Conference on Artificial Intelligence},
  volume = {34},
  number = {04},
  pages = {3187--3194},
  issn = {2374-3468},
  doi = {10.1609/aaai.v34i04.5716},
  url = {https://ojs.aaai.org/index.php/AAAI/article/view/5716},
  abstract = {Spatial interpolation is a class of estimation problems where locations with known values are used to estimate values at other locations, with an emphasis on harnessing spatial locality and trends. Traditional kriging methods have strong Gaussian assumptions, and as a result, often fail to capture complexities within the data. Inspired by the recent progress of graph neural networks, we introduce Kriging Convolutional Networks (KCN), a method of combining advantages of Graph Neural Networks (GNN) and kriging. Compared to standard GNNs, KCNs make direct use of neighboring observations when generating predictions. KCNs also contain the kriging method as a specific configuration. Empirically, we show that this model outperforms GNNs and kriging in several applications.},
  issue = {04},
  langid = {english},
  file = {C:\Users\sjelic\Zotero\storage\AMYZKAG3\Appleby et al. - 2020 - Kriging Convolutional Networks.pdf}
}

@inproceedings{arasSimultaneousOptimizationBerth2014,
  title = {Simultaneous {{Optimization}} of {{Berth Allocation}}, {{Quay Crane Assignment}} and {{Quay Crane Scheduling Problems}} in {{Container Terminals}}},
  booktitle = {Operations {{Research Proceedings}} 2012},
  author = {Aras, Necati and Türkoğulları, Yavuz and Taşkın, Z. Caner and Altınel, Kuban},
  editor = {Helber, Stefan and Breitner, Michael and Rösch, Daniel and Schön, Cornelia and Graf von der Schulenburg, Johann-Matthias and Sibbertsen, Philipp and Steinbach, Marc and Weber, Stefan and Wolter, Anja},
  date = {2014},
  pages = {101--107},
  publisher = {Springer International Publishing},
  location = {Cham},
  doi = {10.1007/978-3-319-00795-3_15},
  abstract = {In this work, we focus on the integrated planning of the following problems faced within the context of seaside operations at container terminals: berth allocation, quay crane assignment, and quay crane scheduling. First, we formulate a new binary integer linear program for the integrated solution of the berth allocation and quay crane assignment problems called BACAP. Then we extend it by incorporating the crane scheduling problem as well, which is named BACASP. Although the model for BACAP is very efficient and even large instances up to 60 vessels can be solved to optimality, only small instances for BACASP can be solved optimally. To be able to solve large instances, we present a necessary and sufficient condition for generating an optimal solution of BACASP from an optimal solution of BACAP using a postprocessing algorithm. We also develop a cutting plane algorithm for the case where this condition is not satisfied. This algorithm solves BACAP repeatedly by adding cuts generated from the optimal solutions at each trial until the aforementioned condition holds.},
  isbn = {978-3-319-00795-3},
  langid = {english},
  file = {C\:\\Users\\sjelic\\Zotero\\storage\\AQC49EF3\\Aras et al. - 2014 - Simultaneous Optimization of Berth Allocation, Qua.pdf;C\:\\Users\\sjelic\\Zotero\\storage\\KNHYSAV5\\1-s2.0-S0167637798000108-main.pdf}
}

@article{aurifeilleBiomimeticApproachMarketing2000,
  title = {A Bio-Mimetic Approach to Marketing Segmentation: {{Principles}} and Comparative Analysis},
  shorttitle = {A Bio-Mimetic Approach to Marketing Segmentation},
  author = {Aurifeille, Jacques-Marie},
  date = {2000},
  journaltitle = {European Journal of Economic and Social Systems},
  shortjournal = {E.J.E.S.S.},
  volume = {14},
  number = {1},
  pages = {93--108},
  issn = {1292-8895, 1292-8909},
  doi = {10.1051/ejess:2000111},
  url = {http://www.edpsciences.org/10.1051/ejess:2000111},
  abstract = {A regression algorithm is proposed for partitioning a population into clusters characterised by homogeneous models and predictions. This algorithm, Typren, is based on the hybridisation of a genetic algorithm with linear regression. An empirical illustration is provided, using real marketing data, which compares Typren with the fuzzy clustering approach, Glimmix, based on a regression mixture model. Typren provides better predictivity and better within-cluster homogeneity of predictions. However, the results are slightly less robust compared to Glimmix.},
  langid = {english},
  file = {C:\Users\sjelic\Zotero\storage\39US6FS6\Aurifeille - 2000 - A bio-mimetic approach to marketing segmentation .pdf}
}

@article{azzariSatelliteMappingTillage2019,
  title = {Satellite Mapping of Tillage Practices in the {{North Central US}} Region from 2005 to 2016},
  author = {Azzari, George and Grassini, Patricio and Edreira, Juan Ignacio Rattalino and Conley, Shawn and Mourtzinis, Spyridon and Lobell, David B.},
  date = {2019-02},
  journaltitle = {Remote Sensing of Environment},
  shortjournal = {Remote Sensing of Environment},
  volume = {221},
  pages = {417--429},
  issn = {00344257},
  doi = {10.1016/j.rse.2018.11.010},
  url = {https://linkinghub.elsevier.com/retrieve/pii/S0034425718305157},
  abstract = {Low-intensity tillage has become more popular among farmers in the United States and many other regions. However, accurate data on when and where low-intensity tillage methods are being used remain scarce, and this scarcity impedes understanding of the factors affecting the adoption and the agronomic or environmental impacts of these practices. In this study, we used composites of satellite imagery from Landsat 5, 7, and 8, and Sentinel-1 in combination with producer data from about 5900 georeferenced fields to train a random forest classifier and generate annual large-scale maps of tillage intensity from 2005 to 2016. We tested different combinations of hyper-parameters using cross-validation, splitting the training and testing data alternatively by field, year, and state to assess the influence of clustering on validation results and evaluate the generalizability of the classification model. We found that the best model was able to map tillage practices across the entire North Central US region at 30 m-resolution with accuracies spanning between 75\% and 79\%, depending on the validation approach. We also found that although Sentinel-1 provides an independent measure that should be sensitive to surface moisture and roughness, it currently adds relatively little to classification performance beyond what is possible with Landsat. When aggregated to the state level, the satellite estimates of percentage lowand high-intensity tillage agreed well with a USDA survey on tillage practices in 2006 (R2 = 0.55). The satellite data also revealed clear increases in low-intensity tillage area for most counties in the past decade. Overall, the ability to accurately map spatial and temporal patterns in tillage should facilitate further study of this important practice in the United States, as well as other regions with fewer survey-based estimates.},
  langid = {english},
  file = {C:\Users\sjelic\Zotero\storage\6K7M3CZG\Azzari et al. - 2019 - Satellite mapping of tillage practices in the Nort.pdf}
}

@article{bagirovAlgorithmClusterwiseLinear2015a,
  title = {An Algorithm for Clusterwise Linear Regression Based on Smoothing Techniques},
  author = {Bagirov, Adil M. and Ugon, Julien and Mirzayeva, Hijran G.},
  date = {2015-02-01},
  journaltitle = {Optimization Letters},
  shortjournal = {Optim Lett},
  volume = {9},
  number = {2},
  pages = {375--390},
  issn = {1862-4480},
  doi = {10.1007/s11590-014-0749-3},
  url = {https://doi.org/10.1007/s11590-014-0749-3},
  abstract = {We propose an algorithm based on an incremental approach and smoothing techniques to solve clusterwise linear regression (CLR) problems. This algorithm incrementally divides the whole data set into groups which can be easily approximated by one linear regression function. A special procedure is introduced to generate an initial solution for solving global optimization problems at each iteration of the incremental algorithm. Such an approach allows one to find global or approximate global solutions to the CLR problems. The algorithm is tested using several data sets for regression analysis and compared with the multistart and incremental Späth algorithms.},
  langid = {english},
  keywords = {Clusterwise regression,Nonconvex optimization,Nonsmooth optimization,Smoothing techniques},
  file = {C:\Users\sjelic\Zotero\storage\ZW5NHNRB\Bagirov и сар. - 2015 - An algorithm for clusterwise linear regression bas.pdf}
}

@article{bagirovDCProgrammingAlgorithm2017,
  title = {{{DC Programming Algorithm}} for {{Clusterwise Linear}} \$\$\{\{\textbackslash varvec\{\vphantom{\}\}\}}{{L}}\vphantom\{\}\vphantom\{\}\vphantom\{\}\_\textbackslash mathbf\{1\}\$\${{Regression}}},
  author = {Bagirov, Adil M. and Taheri, Sona},
  date = {2017-06-01},
  journaltitle = {Journal of the Operations Research Society of China},
  shortjournal = {J. Oper. Res. Soc. China},
  volume = {5},
  number = {2},
  pages = {233--256},
  issn = {2194-6698},
  doi = {10.1007/s40305-017-0151-9},
  url = {https://doi.org/10.1007/s40305-017-0151-9},
  abstract = {The aim of this paper is to develop an algorithm for solving the clusterwise linear least absolute deviations regression problem. This problem is formulated as a nonsmooth nonconvex optimization problem, and the objective function is represented as a difference of convex functions. Optimality conditions are derived by using this representation. An algorithm is designed based on the difference of convex representation and an incremental approach. The proposed algorithm is tested using small to large artificial and real-world data sets.},
  langid = {english},
  keywords = {90C26,90C56,Clusterwise regression,Incremental algorithm,Nonsmooth optimization,Smoothing},
  file = {C:\Users\sjelic\Zotero\storage\58SPHKZA\Bagirov и Taheri - 2017 - DC Programming Algorithm for Clusterwise Linear $$.pdf}
}

@article{bagirovDiscreteGradientMethod2008,
  title = {Discrete {{Gradient Method}}: {{Derivative-Free Method}} for {{Nonsmooth Optimization}}},
  shorttitle = {Discrete {{Gradient Method}}},
  author = {Bagirov, A. M. and Karasözen, B. and Sezer, M.},
  date = {2008-05-01},
  journaltitle = {Journal of Optimization Theory and Applications},
  shortjournal = {J Optim Theory Appl},
  volume = {137},
  number = {2},
  pages = {317--334},
  issn = {1573-2878},
  doi = {10.1007/s10957-007-9335-5},
  url = {https://doi.org/10.1007/s10957-007-9335-5},
  abstract = {A new derivative-free method is developed for solving unconstrained nonsmooth optimization problems. This method is based on the notion of a discrete gradient. It is demonstrated that the discrete gradients can be used to approximate subgradients of a broad class of nonsmooth functions. It is also shown that the discrete gradients can be applied to find descent directions of nonsmooth functions. The preliminary results of numerical experiments with unconstrained nonsmooth optimization problems as well as the comparison of the proposed method with the nonsmooth optimization solver DNLP from CONOPT-GAMS and the derivative-free optimization solver CONDOR are presented.},
  langid = {english},
  keywords = {Derivative-free optimization,Discrete gradients,Nonsmooth optimization,Subdifferentials},
  file = {C:\Users\sjelic\Zotero\storage\L67IYC7C\Bagirov и сар. - 2008 - Discrete Gradient Method Derivative-Free Method f.pdf}
}

@article{bagirovHyperbolicSmoothingFunction2013,
  title = {Hyperbolic Smoothing Function Method for Minimax Problems},
  author = {Bagirov, A.M. and Al Nuaimat, A. and Sultanova, N.},
  date = {2013-06-01},
  journaltitle = {Optimization},
  volume = {62},
  number = {6},
  pages = {759--782},
  publisher = {Taylor \& Francis},
  issn = {0233-1934},
  doi = {10.1080/02331934.2012.675335},
  url = {https://doi.org/10.1080/02331934.2012.675335},
  abstract = {In this article, an approach for solving finite minimax problems is proposed. This approach is based on the use of hyperbolic smoothing functions. In order to apply the hyperbolic smoothing we reformulate the objective function in the minimax problem and study the relationship between the original minimax and reformulated problems. We also study main properties of the hyperbolic smoothing function. Based on these results an algorithm for solving the finite minimax problem is proposed and this algorithm is implemented in general algebraic modelling system. We present preliminary results of numerical experiments with well-known nonsmooth optimization test problems. We also compare the proposed algorithm with the algorithm that uses the exponential smoothing function as well as with the algorithm based on nonlinear programming reformulation of the finite minimax problem.},
  keywords = {65K05,90C25,minimax problem,nonsmooth optimization,smoothing techniques,subdifferential}
}

@article{bagirovIncrementalDCOptimization2021,
  title = {Incremental {{DC}} Optimization Algorithm for Large-Scale Clusterwise Linear Regression},
  author = {Bagirov, Adil M. and Taheri, Sona and Cimen, Emre},
  date = {2021-06-01},
  journaltitle = {Journal of Computational and Applied Mathematics},
  shortjournal = {Journal of Computational and Applied Mathematics},
  volume = {389},
  pages = {113323},
  issn = {0377-0427},
  doi = {10.1016/j.cam.2020.113323},
  url = {https://www.sciencedirect.com/science/article/pii/S0377042720306142},
  abstract = {The objective function in the nonsmooth optimization model of the clusterwise linear regression (CLR) problem with the squared regression error is represented as a difference of two convex functions. Then using the difference of convex algorithm (DCA) approach the CLR problem is replaced by the sequence of smooth unconstrained optimization subproblems. A new algorithm based on the DCA and the incremental approach is designed to solve the CLR problem. We apply the Quasi-Newton method to solve the subproblems. The proposed algorithm is evaluated using several synthetic and real-world data sets for regression and compared with other algorithms for CLR. Results demonstrate that the DCA based algorithm is efficient for solving CLR problems with the large number of data points and in particular, outperforms other algorithms when the number of input variables is small.},
  langid = {english},
  keywords = {Clusterwise linear regression,DC optimization,Nonconvex optimization,Nonsmooth optimization,Regression analysis}
}

@book{bagirovIntroductionNonsmoothOptimization2014,
  title = {Introduction to {{Nonsmooth Optimization}}: {{Theory}}, {{Practice}} and {{Software}}},
  shorttitle = {Introduction to {{Nonsmooth Optimization}}},
  author = {Bagirov, Adil and Karmitsa, Napsu and Mäkelä, Marko M.},
  date = {2014},
  publisher = {Springer International Publishing},
  location = {Cham},
  doi = {10.1007/978-3-319-08114-4},
  url = {https://link.springer.com/10.1007/978-3-319-08114-4},
  isbn = {978-3-319-08113-7 978-3-319-08114-4},
  langid = {english},
  file = {C:\Users\sjelic\Zotero\storage\DILSWTFE\Bagirov и сар. - 2014 - Introduction to Nonsmooth Optimization Theory, Pr.pdf}
}

@article{bagirovModifiedGlobalKmeans2008,
  title = {Modified Global K-Means Algorithm for Minimum Sum-of-Squares Clustering Problems},
  author = {Bagirov, Adil M.},
  date = {2008-10-01},
  journaltitle = {Pattern Recognition},
  shortjournal = {Pattern Recognition},
  volume = {41},
  number = {10},
  pages = {3192--3199},
  issn = {0031-3203},
  doi = {10.1016/j.patcog.2008.04.004},
  url = {https://www.sciencedirect.com/science/article/pii/S0031320308001362},
  abstract = {k-Means algorithm and its variations are known to be fast clustering algorithms. However, they are sensitive to the choice of starting points and inefficient for solving clustering problems in large data sets. Recently, a new version of the k-means algorithm, the global k-means algorithm has been developed. It is an incremental algorithm that dynamically adds one cluster center at a time and uses each data point as a candidate for the k-th cluster center. Results of numerical experiments show that the global k-means algorithm considerably outperforms the k-means algorithms. In this paper, a new version of the global k-means algorithm is proposed. A starting point for the k-th cluster center in this algorithm is computed by minimizing an auxiliary cluster function. Results of numerical experiments on 14 data sets demonstrate the superiority of the new algorithm, however, it requires more computational time than the global k-means algorithm.},
  langid = {english},
  keywords = {-Means algorithm,Global -means algorithm,Minimum sum-of-squares clustering,Nonsmooth optimization},
  file = {C\:\\Users\\sjelic\\Zotero\\storage\\PITFVXI3\\Bagirov - 2008 - Modified global k-means algorithm for minimum sum-.pdf;C\:\\Users\\sjelic\\Zotero\\storage\\QHMRU9H7\\S0031320308001362.html}
}

@article{bagirovNonsmoothDCProgramming2018,
  title = {Nonsmooth {{DC}} Programming Approach to Clusterwise Linear Regression: Optimality Conditions and Algorithms},
  shorttitle = {Nonsmooth {{DC}} Programming Approach to Clusterwise Linear Regression},
  author = {Bagirov, A.M. and Ugon, J.},
  date = {2018-01-02},
  journaltitle = {Optimization Methods and Software},
  volume = {33},
  number = {1},
  pages = {194--219},
  publisher = {Taylor \& Francis},
  issn = {1055-6788},
  doi = {10.1080/10556788.2017.1371717},
  url = {https://doi.org/10.1080/10556788.2017.1371717},
  abstract = {The clusterwise linear regression problem is formulated as a nonsmooth nonconvex optimization problem using the squared regression error function. The objective function in this problem is represented as a difference of convex functions. Optimality conditions are derived, and an algorithm is designed based on such a representation. An incremental approach is proposed to generate starting solutions. The algorithm is tested on small to large data sets.},
  keywords = {90C26,90C56,cluster analysis,DC programming,nonsmooth optimization,regression analysis}
}

@article{bagirovNonsmoothNonconvexOptimization2013,
  title = {Nonsmooth Nonconvex Optimization Approach to Clusterwise Linear Regression Problems},
  author = {Bagirov, Adil M. and Ugon, Julien and Mirzayeva, Hijran},
  date = {2013-08-16},
  journaltitle = {European Journal of Operational Research},
  shortjournal = {European Journal of Operational Research},
  volume = {229},
  number = {1},
  pages = {132--142},
  issn = {0377-2217},
  doi = {10.1016/j.ejor.2013.02.059},
  url = {https://www.sciencedirect.com/science/article/pii/S0377221713002087},
  abstract = {Clusterwise regression consists of finding a number of regression functions each approximating a subset of the data. In this paper, a new approach for solving the clusterwise linear regression problems is proposed based on a nonsmooth nonconvex formulation. We present an algorithm for minimizing this nonsmooth nonconvex function. This algorithm incrementally divides the whole data set into groups which can be easily approximated by one linear regression function. A special procedure is introduced to generate a good starting point for solving global optimization problems at each iteration of the incremental algorithm. Such an approach allows one to find global or near global solution to the problem when the data sets are sufficiently dense. The algorithm is compared with the multistart Späth algorithm on several publicly available data sets for regression analysis.},
  langid = {english},
  keywords = {Clusterwise linear regression,Incremental algorithm,Späth algorithm},
  file = {C:\Users\sjelic\Zotero\storage\NJU4KHW7\Bagirov et al. - 2013 - Nonsmooth nonconvex optimization approach to clust.pdf}
}

@article{bagirovNonsmoothOptimizationAlgorithm2015,
  title = {Nonsmooth {{Optimization Algorithm}} for {{Solving Clusterwise Linear Regression Problems}}},
  author = {Bagirov, Adil M. and Ugon, Julien and Mirzayeva, Hijran G.},
  date = {2015-03-01},
  journaltitle = {Journal of Optimization Theory and Applications},
  shortjournal = {J Optim Theory Appl},
  volume = {164},
  number = {3},
  pages = {755--780},
  issn = {1573-2878},
  doi = {10.1007/s10957-014-0566-y},
  url = {https://doi.org/10.1007/s10957-014-0566-y},
  abstract = {Clusterwise linear regression consists of finding a number of linear regression functions each approximating a subset of the data. In this paper, the clusterwise linear regression problem is formulated as a nonsmooth nonconvex optimization problem and an algorithm based on an incremental approach and on the discrete gradient method of nonsmooth optimization is designed to solve it. This algorithm incrementally divides the whole dataset into groups which can be easily approximated by one linear regression function. A special procedure is introduced to generate good starting points for solving global optimization problems at each iteration of the incremental algorithm. The algorithm is compared with the multi-start Späth and the incremental algorithms on several publicly available datasets for regression analysis.},
  langid = {english},
  keywords = {65K05,90C25,Clusterwise linear regression,Discrete gradient method,Nonconvex optimization,Nonsmooth optimization},
  file = {C:\Users\sjelic\Zotero\storage\VF52JVLZ\Bagirov и сар. - 2015 - Nonsmooth Optimization Algorithm for Solving Clust.pdf}
}

@article{bagirovPredictionMonthlyRainfall2017,
  title = {Prediction of Monthly Rainfall in {{Victoria}}, {{Australia}}: {{Clusterwise}} Linear Regression~Approach},
  shorttitle = {Prediction of Monthly Rainfall in {{Victoria}}, {{Australia}}},
  author = {Bagirov, Adil M. and Mahmood, Arshad and Barton, Andrew},
  date = {2017-05-15},
  journaltitle = {Atmospheric Research},
  shortjournal = {Atmospheric Research},
  volume = {188},
  pages = {20--29},
  issn = {0169-8095},
  doi = {10.1016/j.atmosres.2017.01.003},
  url = {https://www.sciencedirect.com/science/article/pii/S0169809517300285},
  abstract = {This paper develops the Clusterwise Linear Regression (CLR) technique for prediction of monthly rainfall. The CLR is a combination of clustering and regression techniques. It is formulated as an optimization problem and an incremental algorithm is designed to solve it. The algorithm is applied to predict monthly rainfall in Victoria, Australia using rainfall data with five input meteorological variables over the period of 1889–2014 from eight geographically diverse weather stations. The prediction performance of the CLR method is evaluated by comparing observed and predicted rainfall values using four measures of forecast accuracy. The proposed method is also compared with the CLR using the maximum likelihood framework by the expectation-maximization algorithm, multiple linear regression, artificial neural networks and the support vector machines for regression models using computational results. The results demonstrate that the proposed algorithm outperforms other methods in most locations.},
  langid = {english},
  keywords = {Cluster analysis,Clusterwise linear regression.,Prediction models,Rainfall prediction,Regression analysis},
  file = {C\:\\Users\\sjelic\\Zotero\\storage\\SCFXHF83\\Bagirov et al. - 2017 - Prediction of monthly rainfall in Victoria, Austra.pdf;C\:\\Users\\sjelic\\Zotero\\storage\\ZI4CVL7C\\S0169809517300285.html}
}

@online{bahdanauEndtoEndAttentionbasedLarge2016,
  title = {End-to-{{End Attention-based Large Vocabulary Speech Recognition}}},
  author = {Bahdanau, Dzmitry and Chorowski, Jan and Serdyuk, Dmitriy and Brakel, Philemon and Bengio, Yoshua},
  date = {2016-03-14},
  eprint = {1508.04395},
  eprinttype = {arXiv},
  eprintclass = {cs},
  doi = {10.48550/arXiv.1508.04395},
  url = {http://arxiv.org/abs/1508.04395},
  abstract = {Many of the current state-of-the-art Large Vocabulary Continuous Speech Recognition Systems (LVCSR) are hybrids of neural networks and Hidden Markov Models (HMMs). Most of these systems contain separate components that deal with the acoustic modelling, language modelling and sequence decoding. We investigate a more direct approach in which the HMM is replaced with a Recurrent Neural Network (RNN) that performs sequence prediction directly at the character level. Alignment between the input features and the desired character sequence is learned automatically by an attention mechanism built into the RNN. For each predicted character, the attention mechanism scans the input sequence and chooses relevant frames. We propose two methods to speed up this operation: limiting the scan to a subset of most promising frames and pooling over time the information contained in neighboring frames, thereby reducing source sequence length. Integrating an n-gram language model into the decoding process yields recognition accuracies similar to other HMM-free RNN-based approaches.},
  pubstate = {prepublished},
  keywords = {Computer Science - Artificial Intelligence,Computer Science - Computation and Language,Computer Science - Machine Learning,Computer Science - Neural and Evolutionary Computing},
  file = {C\:\\Users\\sjelic\\Zotero\\storage\\DK8FTNAA\\Bahdanau et al. - 2016 - End-to-End Attention-based Large Vocabulary Speech.pdf;C\:\\Users\\sjelic\\Zotero\\storage\\QFD8552F\\1508.html}
}

@article{baiEstimationSoilOrganic2022,
  title = {Estimation of {{Soil Organic Carbon Using Vis-NIR Spectral Data}} and {{Spectral Feature Bands Selection}} in {{Southern Xinjiang}}, {{China}}},
  author = {Bai, Zijin and Xie, Modong and Hu, Bifeng and Luo, Defang and Wan, Chang and Peng, Jie and Shi, Zhou},
  date = {2022-01},
  journaltitle = {Sensors},
  volume = {22},
  number = {16},
  pages = {6124},
  publisher = {Multidisciplinary Digital Publishing Institute},
  issn = {1424-8220},
  doi = {10.3390/s22166124},
  url = {https://www.mdpi.com/1424-8220/22/16/6124},
  abstract = {Soil organic carbon (SOC) plays an important role in the global carbon cycle and soil fertility supply. Rapid and accurate estimation of SOC content could provide critical information for crop production, soil management and soil carbon pool regulation. Many researchers have confirmed the feasibility and great potential of visible and near-infrared (Vis-NIR) spectroscopy in evaluating SOC content rapidly and accurately. Here, to evaluate the feasibility of different spectral bands variable selection methods for SOC prediction, we collected a total of 330 surface soil samples from the cotton field in the Alar Reclamation area in the southern part of Xinjiang, which is located in the arid region of northwest China. Then, we estimated the SOC content using laboratory Vis-NIR spectral. The Particle Swarm optimization (PSO), Competitive adaptive reweighted sampling (CARS) and Ant colony optimization (ACO) were adopted to select SOC feature bands. The partial least squares regression (PLSR), random forest (RF) and convolutional neural network (CNN) inversion models were constructed by using full-bands (400–2400 nm) spectra (R) and feature bands, respectively. And we also analyzed the effects of spectral feature band selection methods and modeling methods on the prediction accuracy of SOC. The results indicated that: (1) There are significant differences in the feature bands selected using different methods. The feature bands selected methods substantially reduced the spectral variable dimensionality and model complexity. The models built by the feature bands selected by CARS, PSO and ACO methods showed the different potential of improvement in model accuracy compared with the full-band models. (2) The CNN model had the best performance for predicting SOC. The R2 of the optimal CNN model is 0.90 in the validation, which was improved by 0.05 and 0.04 in comparison with the PLSR and RF model, respectively. (3) The highest prediction accuracy was archived by the CNN model using the feature bands selected by CARS (validation set R2 = 0.90, RMSE = 0.97 g kg−1, RPD = 3.18, RPIQ = 3.11). This study indicated that using the CARS method to select spectral feature bands, combined with the CNN modeling method can well predict SOC content with higher accuracy.},
  issue = {16},
  langid = {english},
  keywords = {deep learning,SOC,spectral feature bands selection,Vis-NIR spectroscopy},
  file = {C:\Users\sjelic\Zotero\storage\4CTNDUDS\Bai et al. - 2022 - Estimation of Soil Organic Carbon Using Vis-NIR Sp.pdf}
}

@thesis{bascouSparseLinearModel2022,
  type = {phdthesis},
  title = {Sparse Linear Model with Quadratic Interactions},
  author = {Bascou, Florent},
  date = {2022-09-09},
  institution = {Université de Montpellier},
  url = {https://theses.hal.science/tel-04058087},
  abstract = {We present an estimator for the high-dimensional fitting of a linear model with quadratic interactions. As such a model has a very large number of features, its estimation raises many statistical and computational challenges. Thus, its estimation has motivated a lot of work over the last two decades, and remains a challenge in many applications. From a statistical point of view, one of the challenges is to be able to select the features, to facilitate the interpretability of the model. Moreover, since the added interaction features can be highly correlated, an adapted regularization must be able to take them into account. We then propose to adapt the Elastic Net estimator, to take into account the potential correlations thanks to the l2 penalty, and to obtain a parsimonious model using the l1 penalty. Moreover, a common approach used in the literature, to favor main effects while reducing the number of interactions to be considered, is the heredity assumption. This assumption allows the inclusion of an interaction only if and when the associated main effects are selected in the model. Thus, it leads to parsimonious models, easier to interpret, while reducing the number of interactions to be visited and the computational cost. However, it does not allow the exploration of interaction variables whose main effects are not selected, although these variables may be relevant to consider. We therefore propose to emancipate ourselves from this structural heredity assumption, and to penalize interactions more than main effects, in order to favor the latter and interpretability. It is also known that penalized estimators such as Elastic Net bias the coefficients by ag- gressively shrinking them towards zero. A consequence is the selection of additional features to compensate for the loss of amplitude of the penalized coefficients, which affects the calibration of the hyperparameters during cross-validation. A simple solution is then to select the features by the Elastic Net, then to estimate these coefficients by the Least Squares estimator, for each hyperparameter. However, if the features are highly correlated, the Least Squares step may fail. Therefore, we choose to adapt a debiasing method allowing to obtain simultaneously the Elastic Net coefficients and their debiased version. A first challenge of this work is to develop an algorithm that does not require to store the interaction matrix, which could exceed the memory capacity of a computer. To do this, we adapt a coordinate descent algorithm, allowing to build the columns of this matrix on- the-fly. Although this step avoids storage, it adds extra computations to each step of the algorithm, thus increasing its computation time. Moreover, knowing that our estimator is parsimonious, these computations may be all the more useless as many interaction coefficients are zero, and thus unnecessarily updated. A second issue is then to propose an algorithm that remains computationally efficient, despite the large number of interactions to consider and this computational overhead. Therefore, to exploit the parsimony of the estimator and to reduce the number of interaction coefficients to be updated, we adapt an active set algorithm. Second, we adapt the Anderson acceleration, which allows us to speed up the coordinate descent algorithms for solving LASSO type problems. Finally, the performance of our estimator is illustrated on simulated and real data, and compared with state-of-the-art methods.},
  langid = {english},
  file = {C:\Users\sjelic\Zotero\storage\XWFED2I5\Bascou - 2022 - Sparse linear model with quadratic interactions.pdf}
}

@article{bazirhaNovelMILPFormulation2023,
  title = {A Novel {{MILP}} Formulation and an Efficient Heuristic for the Vehicle Routing Problem with Lunch Break},
  author = {Bazirha, Mohammed},
  date = {2023-12-08},
  journaltitle = {Annals of Operations Research},
  shortjournal = {Ann Oper Res},
  issn = {1572-9338},
  doi = {10.1007/s10479-023-05742-3},
  url = {https://doi.org/10.1007/s10479-023-05742-3},
  abstract = {The vehicle routing problem with time windows and lunch break (VRPTW-LB) is an NP-hard combinatorial optimization problem that belongs to the vehicle routing problem family. In the VRPTW-LB, each vehicle must serve assigned customers within their availability periods and take a mandatory break within its time slot for a given duration. This paper proposes a new mixed integer linear programming (MILP) formulation for the VRPTW-LB. The new MILP formulation is based on the idea of the intersection of two intervals, which must be non-empty. It provides a flexible schedule for lunch breaks by defining the earliest and latest start times at which each break can be taken, instead of defining its exact start time. A comparative study of MILP formulations from the literature and the proposed one is performed, which are tested on benchmark instances from the literature. These MILP formulations are powerless to solve large-scale instances. To overcome this limitation, an efficient lunch break scheduling algorithm is proposed and embedded into a simulated annealing (SA) based heuristic. Computational results highlight the competitiveness of the new MILP formulation with respect to other MILP formulations and the efficiency of the proposed heuristic in obtaining high-quality solutions in short CPU running times.},
  langid = {english},
  keywords = {Continuous Optimization,Logistics,Lunch break,Mathematical modeling,Milgram Experiment,Operations Research and Decision Theory,Optimization,Simulated annealing,Time windows,Transportation Technology and Traffic Engineering,Vehicle routing problem},
  file = {C:\Users\sjelic\Zotero\storage\757L7WSR\Bazirha - 2023 - A novel MILP formulation and an efficient heuristic for the vehicle routing problem with lunch break.pdf}
}

@article{beheshtiVehicleRoutingProblem2015,
  title = {The Vehicle Routing Problem with Multiple Prioritized Time Windows: {{A}} Case Study},
  shorttitle = {The Vehicle Routing Problem with Multiple Prioritized Time Windows},
  author = {Beheshti, Ali Kourank and Hejazi, Seyed Reza and Alinaghian, Mehdi},
  date = {2015-12-01},
  journaltitle = {Computers \& Industrial Engineering},
  shortjournal = {Computers \& Industrial Engineering},
  volume = {90},
  pages = {402--413},
  issn = {0360-8352},
  doi = {10.1016/j.cie.2015.10.005},
  url = {https://www.sciencedirect.com/science/article/pii/S0360835215003988},
  abstract = {This paper addresses Multi-objective Vehicle Routing Problem with Multiple Prioritized Time Windows (VRPMPTW) in which the distributer proposes a set of all non-overlapping time windows with equal or different lengths and the customers prioritize these delivery time windows. VRPMPTW aims to find a set of routes of minimal total traveling cost and maximal customer satisfaction (with regard to the prioritized time windows), starting and ending at the depot, in such a way that each customer is visited by one vehicle given the capacity of the vehicle to satisfy a specific demand. This problem is inspired from a real life application. The contribution of this paper lies in its addressing the VRPMPTW from a problem definition, modeling and methodological point of view. We developed a mathematical model for this problem. This model can simply be used for a wide range of applications where the customers have multiple flexible time windows and violation of time windows may drop the satisfaction levels of customers and lead to profit loss in the long term. A Cooperative Coevolutionary Multi-objective Quantum-Genetic Algorithm (CCMQGA) is also proposed to solve this problem. A new local search is designed and used in CCMQGA to reach an appropriate pareto front. Finally, the proposed approach is employed in a real case study and the results of the proposed CCMQGA are compared with the current solution obtained from managerial experience, the results of NSGA-II and the multi-objective quantum-inspired evolutionary algorithm.},
  keywords = {Cooperative coevolutionary multi-objective quantum-genetic algorithm,Multi-objective optimization,Vehicle routing problem with multiple prioritized time windows},
  file = {C\:\\Users\\sjelic\\Zotero\\storage\\YUA7IGTZ\\Beheshti et al. - 2015 - The vehicle routing problem with multiple prioritized time windows A case study.pdf;C\:\\Users\\sjelic\\Zotero\\storage\\3WLEUGZ6\\S0360835215003988.html}
}

@article{belhaizaHybridVariableNeighborhood2014,
  title = {A Hybrid Variable Neighborhood Tabu Search Heuristic for the Vehicle Routing Problem with Multiple Time Windows},
  author = {Belhaiza, Slim and Hansen, Pierre and Laporte, Gilbert},
  date = {2014-12-01},
  journaltitle = {Computers \& Operations Research},
  shortjournal = {Computers \& Operations Research},
  series = {Recent Advances in {{Variable}} Neighborhood Search},
  volume = {52},
  pages = {269--281},
  issn = {0305-0548},
  doi = {10.1016/j.cor.2013.08.010},
  url = {https://www.sciencedirect.com/science/article/pii/S0305054813002165},
  abstract = {This paper presents a new hybrid variable neighborhood-tabu search heuristic for the Vehicle Routing Problem with Multiple Time windows. It also proposes a minimum backward time slack algorithm applicable to a multiple time windows environment. This algorithm records the minimum waiting time and the minimum delay during route generation and adjusts the arrival and departure times backward. The implementation of the proposed heuristic is compared to an ant colony heuristic on benchmark instances involving multiple time windows. Computational results on newly generated instances are provided.},
  keywords = {Multiple time windows,Tabu search,Variable neighborhood search,Vehicle routing problem},
  file = {C\:\\Users\\sjelic\\Zotero\\storage\\U7HKIZBM\\Belhaiza et al. - 2014 - A hybrid variable neighborhood tabu search heuristic for the vehicle routing problem with multiple t.pdf;C\:\\Users\\sjelic\\Zotero\\storage\\M7DQX4PI\\S0305054813002165.html}
}

@article{ben-davidTheoryLearningDifferent2010,
  title = {A Theory of Learning from Different Domains},
  author = {Ben-David, Shai and Blitzer, John and Crammer, Koby and Kulesza, Alex and Pereira, Fernando and Vaughan, Jennifer Wortman},
  date = {2010-05-01},
  journaltitle = {Machine Learning},
  shortjournal = {Mach Learn},
  volume = {79},
  number = {1},
  pages = {151--175},
  issn = {1573-0565},
  doi = {10.1007/s10994-009-5152-4},
  url = {https://doi.org/10.1007/s10994-009-5152-4},
  abstract = {Discriminative learning methods for classification perform well when training and test data are drawn from the same distribution. Often, however, we have plentiful labeled training data from a source domain but wish to learn a classifier which performs well on a target domain with a different distribution and little or no labeled training data. In this work we investigate two questions. First, under what conditions can a classifier trained from source data be expected to perform well on target data? Second, given a small amount of labeled target data, how should we combine it during training with the large amount of labeled source data to achieve the lowest target error at test time?},
  langid = {english},
  file = {C:\Users\sjelic\Zotero\storage\3NVR6KWU\Ben-David et al. - 2010 - A theory of learning from different domains.pdf}
}

@article{benahmedEffectDifferentMathematical2023,
  title = {The Effect of Different Mathematical Formulations on a Matheuristic Algorithm for the Production Routing Problem},
  author = {Ben Ahmed, Mohamed and Hvattum, Lars Magnus and Agra, Agostinho},
  date = {2023-07-01},
  journaltitle = {Computers \& Operations Research},
  shortjournal = {Computers \& Operations Research},
  volume = {155},
  pages = {106232},
  issn = {0305-0548},
  doi = {10.1016/j.cor.2023.106232},
  url = {https://www.sciencedirect.com/science/article/pii/S0305054823000965},
  abstract = {We perform an experimental study to evaluate the performance of a matheuristic for the production routing problem (PRP). First, we develop a basic matheuristic that prescribes starting from a partial initial solution, completing it using a sequence of constructive heuristics, and improving it using a general-purpose mixed-integer programming heuristic. Next, we investigate the effect of three state-of-the-art mathematical formulations on the proposed matheuristic convergence. The formulations are implemented and tested with and without the use of valid inequalities. In addition, by suggesting different techniques to generate a feasible starting solution for our matheuristic, we assess the contribution of an initial solution to the matheuristic’s overall performance. We conduct extensive computational experiments on benchmark data instances for the PRP. The results show that a proper choice of an embedded mathematical formulation depends on the data instances’ features, such as the number of customers and the length of the planning horizon. The comparisons undertaken in this study indicate that having a better initial solution does not necessarily lead to finding a better final solution.},
  keywords = {Experimental analysis,Inventory routing,Lot sizing,Mixed integer formulation,Proximity search,Valid inequality},
  file = {C:\Users\sjelic\Zotero\storage\SIATIJT9\S0305054823000965.html}
}

@article{bendersPartitioningProceduresSolving1962,
  title = {Partitioning procedures for solving mixed-variables programming problems.},
  author = {Benders, J. F.},
  date = {1962/0063},
  journaltitle = {Numerische Mathematik},
  volume = {4},
  pages = {238--252},
  issn = {0029-599X; 0945-3245/e},
  url = {https://eudml.org/doc/urn:eudml:doc:131533},
  langid = {nil},
  file = {C:\Users\sjelic\Zotero\storage\5NC6FT2V\urneudmldoc131533.html}
}

@article{berkValidPostselectionInference2013,
  title = {Valid Post-Selection Inference},
  author = {Berk, Richard and Brown, Lawrence and Buja, Andreas and Zhang, Kai and Zhao, Linda},
  date = {2013-04-01},
  journaltitle = {The Annals of Statistics},
  shortjournal = {Ann. Statist.},
  volume = {41},
  number = {2},
  issn = {0090-5364},
  doi = {10.1214/12-AOS1077},
  url = {https://projecteuclid.org/journals/annals-of-statistics/volume-41/issue-2/Valid-post-selection-inference/10.1214/12-AOS1077.full},
  langid = {english},
  file = {C:\Users\sjelic\Zotero\storage\6AMXBM8B\Berk et al. - 2013 - Valid post-selection inference.pdf}
}

@article{bertsimasClassificationRegressionInteger2007,
  title = {Classification and {{Regression}} via {{Integer Optimization}}},
  author = {Bertsimas, Dimitris and Shioda, Romy},
  date = {2007-04},
  journaltitle = {Operations Research},
  volume = {55},
  number = {2},
  pages = {252--271},
  publisher = {INFORMS},
  issn = {0030-364X},
  doi = {10.1287/opre.1060.0360},
  url = {https://pubsonline.informs.org/doi/10.1287/opre.1060.0360},
  abstract = {Motivated by the significant advances in integer optimization in the past decade, we introduce mixed-integer optimization methods to the classical statistical problems of classification and regression and construct a software package called CRIO (classification and regression via integer optimization). CRIO separates data points into different polyhedral regions. In classification each region is assigned a class, while in regression each region has its own distinct regression coefficients. Computational experimentations with generated and real data sets show that CRIO is comparable to and often outperforms the current leading methods in classification and regression. We hope that these results illustrate the potential for significant impact of integer optimization methods on computational statistics and data mining.},
  keywords = {applications,integer,nonparametric,programming,statistics},
  file = {C:\Users\sjelic\Zotero\storage\3YRXFJCZ\Bertsimas и Shioda - 2007 - Classification and Regression via Integer Optimiza.pdf}
}

@article{bienLassoHierarchicalInteractions2013a,
  title = {A Lasso for Hierarchical Interactions},
  author = {Bien, Jacob and Taylor, Jonathan and Tibshirani, Robert},
  date = {2013-06},
  journaltitle = {The Annals of Statistics},
  volume = {41},
  number = {3},
  pages = {1111--1141},
  publisher = {Institute of Mathematical Statistics},
  issn = {0090-5364, 2168-8966},
  doi = {10.1214/13-AOS1096},
  url = {https://projecteuclid.org/journals/annals-of-statistics/volume-41/issue-3/A-lasso-for-hierarchical-interactions/10.1214/13-AOS1096.full},
  abstract = {We add a set of convex constraints to the lasso to produce sparse interaction models that honor the hierarchy restriction that an interaction only be included in a model if one or both variables are marginally important. We give a precise characterization of the effect of this hierarchy constraint, prove that hierarchy holds with probability one and derive an unbiased estimate for the degrees of freedom of our estimator. A bound on this estimate reveals the amount of fitting “saved” by the hierarchy constraint. We distinguish between parameter sparsity—the number of nonzero coefficients—and practical sparsity—the number of raw variables one must measure to make a new prediction. Hierarchy focuses on the latter, which is more closely tied to important data collection concerns such as cost, time and effort. We develop an algorithm, available in the R package hierNet, and perform an empirical study of our method.},
  keywords = {62J07,convexity,hierarchical sparsity,interactions,Lasso,regularized regression},
  file = {C:\Users\sjelic\Zotero\storage\RDHAMPMW\Bien et al. - 2013 - A lasso for hierarchical interactions.pdf}
}

@article{bierwirthFollowsurveyBerthAllocation2015,
  title = {A Follow-up Survey of Berth Allocation and Quay Crane Scheduling Problems in Container Terminals},
  author = {Bierwirth, Christian and Meisel, Frank},
  date = {2015-08-01},
  journaltitle = {European Journal of Operational Research},
  shortjournal = {European Journal of Operational Research},
  volume = {244},
  number = {3},
  pages = {675--689},
  issn = {0377-2217},
  doi = {10.1016/j.ejor.2014.12.030},
  url = {https://www.sciencedirect.com/science/article/pii/S0377221714010480},
  abstract = {This paper surveys recent publications on berth allocation, quay crane assignment, and quay crane scheduling problems in seaport container terminals. It continues the survey of Bierwirth and Meisel (2010) that covered the research up to 2009. Since then, there was a strong increase of activity observed in this research field resulting in more than 120 new publications. In this paper, we classify this new literature according to the features of models considered for berth allocation, quay crane scheduling and integrated approaches by using the classification schemes proposed in the preceding survey. Moreover, we identify trends in the field, we take a look at the methods that have been developed for solving new models, we discuss ways for evaluating models and algorithms, and, finally, we light up potential directions for future research.},
  keywords = {Berth allocation,Integrated approaches,Quay crane assignment,Quay crane scheduling},
  file = {C\:\\Users\\sjelic\\Zotero\\storage\\NLIGFNYA\\Bierwirth and Meisel - 2015 - A follow-up survey of berth allocation and quay cr.pdf;C\:\\Users\\sjelic\\Zotero\\storage\\FB855KI8\\S0377221714010480.html}
}

@article{bierwirthSurveyBerthAllocation2010,
  title = {A Survey of Berth Allocation and Quay Crane Scheduling Problems in Container Terminals},
  author = {Bierwirth, Christian and Meisel, Frank},
  date = {2010-05-01},
  journaltitle = {European Journal of Operational Research},
  shortjournal = {European Journal of Operational Research},
  volume = {202},
  number = {3},
  pages = {615--627},
  issn = {0377-2217},
  doi = {10.1016/j.ejor.2009.05.031},
  url = {https://www.sciencedirect.com/science/article/pii/S0377221709003579},
  abstract = {Due to the variety of technical equipments and terminal layouts, research has produced a multitude of optimization models for seaside operations planning in container terminals. To provide a support in modeling problem characteristics and in suggesting applicable algorithms this paper reviews the relevant literature. For this purpose new classification schemes for berth allocation problems and quay crane scheduling problems are developed. Particular focus is put on integrated solution approaches which receive increasing importance for the terminal management.},
  keywords = {Berth allocation,Container terminal operations,Integrated planning,Problem classification,Quay crane assignment,Quay crane scheduling},
  file = {C\:\\Users\\sjelic\\Zotero\\storage\\5TV2N2N3\\Bierwirth and Meisel - 2010 - A survey of berth allocation and quay crane schedu.pdf;C\:\\Users\\sjelic\\Zotero\\storage\\3R8DGWVM\\S0377221709003579.html}
}

@article{bischlHyperparameterOptimizationFoundations2023,
  title = {Hyperparameter Optimization: {{Foundations}}, Algorithms, Best Practices, and Open Challenges},
  shorttitle = {Hyperparameter Optimization},
  author = {Bischl, Bernd and Binder, Martin and Lang, Michel and Pielok, Tobias and Richter, Jakob and Coors, Stefan and Thomas, Janek and Ullmann, Theresa and Becker, Marc and Boulesteix, Anne-Laure and Deng, Difan and Lindauer, Marius},
  date = {2023},
  journaltitle = {WIREs Data Mining and Knowledge Discovery},
  volume = {13},
  number = {2},
  pages = {e1484},
  issn = {1942-4795},
  doi = {10.1002/widm.1484},
  url = {https://onlinelibrary.wiley.com/doi/abs/10.1002/widm.1484},
  abstract = {Most machine learning algorithms are configured by a set of hyperparameters whose values must be carefully chosen and which often considerably impact performance. To avoid a time-consuming and irreproducible manual process of trial-and-error to find well-performing hyperparameter configurations, various automatic hyperparameter optimization (HPO) methods—for example, based on resampling error estimation for supervised machine learning—can be employed. After introducing HPO from a general perspective, this paper reviews important HPO methods, from simple techniques such as grid or random search to more advanced methods like evolution strategies, Bayesian optimization, Hyperband, and racing. This work gives practical recommendations regarding important choices to be made when conducting HPO, including the HPO algorithms themselves, performance evaluation, how to combine HPO with machine learning pipelines, runtime improvements, and parallelization. This article is categorized under: Algorithmic Development {$>$} Statistics Technologies {$>$} Machine Learning Technologies {$>$} Prediction},
  langid = {english},
  keywords = {automl,hyperparameter optimization,machine learning,model selection,tuning},
  file = {C\:\\Users\\sjelic\\Zotero\\storage\\PVBXHDG2\\Bischl et al. - 2023 - Hyperparameter optimization Foundations, algorith.pdf;C\:\\Users\\sjelic\\Zotero\\storage\\3IEPHZPT\\widm.html}
}

@article{boschettiMatheuristicsUsingMathematics2022,
  title = {Matheuristics: Using Mathematics for Heuristic Design},
  shorttitle = {Matheuristics},
  author = {Boschetti, Marco Antonio and Maniezzo, Vittorio},
  date = {2022-06-01},
  journaltitle = {4OR},
  shortjournal = {4OR-Q J Oper Res},
  volume = {20},
  number = {2},
  pages = {173--208},
  issn = {1614-2411},
  doi = {10.1007/s10288-022-00510-8},
  url = {https://doi.org/10.1007/s10288-022-00510-8},
  abstract = {Matheuristics are heuristic algorithms based on mathematical tools such as the ones provided by mathematical programming, that are structurally general enough to be applied to different problems with little adaptations to their abstract structure. The result can be metaheuristic hybrids having components derived from the mathematical model of the problems of interest, but the mathematical techniques themselves can define general heuristic solution frameworks. In this paper, we focus our attention on mathematical programming and its contributions to developing effective heuristics. We briefly describe the mathematical tools available and then some matheuristic approaches, reporting some representative examples from the literature. We also take the opportunity to provide some ideas for possible future development.},
  langid = {english},
  keywords = {90-00,90-02,90C11,Heuristics,Mathematical programming,Matheuristics},
  file = {C:\Users\sjelic\Zotero\storage\8DCWCCS3\Boschetti and Maniezzo - 2022 - Matheuristics using mathematics for heuristic des.pdf}
}

@article{breimanRandomForests2001,
  title = {Random {{Forests}}},
  author = {Breiman, Leo},
  date = {2001-10-01},
  journaltitle = {Mach. Learn.},
  volume = {45},
  number = {1},
  pages = {5--32},
  issn = {0885-6125},
  doi = {10.1023/A:1010933404324},
  url = {https://doi.org/10.1023/A:1010933404324},
  abstract = {Random forests are a combination of tree predictors such that each tree depends on the values of a random vector sampled independently and with the same distribution for all trees in the forest. The generalization error for forests converges a.s. to a limit as the number of trees in the forest becomes large. The generalization error of a forest of tree classifiers depends on the strength of the individual trees in the forest and the correlation between them. Using a random selection of features to split each node yields error rates that compare favorably to Adaboost (Y. Freund \& R. Schapire, Machine Learning: Proceedings of the Thirteenth International conference, ***, 148–156), but are more robust with respect to noise. Internal estimates monitor error, strength, and correlation and these are used to show the response to increasing the number of features used in the splitting. Internal estimates are also used to measure variable importance. These ideas are also applicable to regression.},
  file = {C:\Users\sjelic\Zotero\storage\FCT57UG8\Breiman - 2001 - Random Forests.pdf}
}

@article{bruscoSimulatedAnnealingHeuristic2002,
  title = {A {{Simulated Annealing Heuristic}} for a {{Bicriterion Partitioning Problem}} in {{Market Segmentation}}},
  author = {Brusco, Michael J. and Cradit, J. Dennis and Stahl, Stephanie},
  date = {2002-02-01},
  journaltitle = {Journal of Marketing Research},
  volume = {39},
  number = {1},
  pages = {99--109},
  publisher = {SAGE Publications Inc},
  issn = {0022-2437},
  doi = {10.1509/jmkr.39.1.99.18932},
  url = {https://doi.org/10.1509/jmkr.39.1.99.18932},
  abstract = {K-means clustering procedures are frequently used to identify homogeneous market segments on the basis of a set of descriptor variables. In practice, however, market research analysts often desire both homogeneous market segments and good explanation of an exogenous response variable. Unfortunately, the relationship between these two objective criteria can be antagonistic, and it is often difficult to find clustering solutions that yield adequate levels for both criteria. The authors present a simulated annealing heuristic for solving bicriterion partitioning problems related to these objectives. A large computational study and an empirical demonstration reveal the effectiveness of the methodology. The authors also discuss limitations and extensions of the method.},
  langid = {english}
}

@article{caporossiVariableNeighborhoodSearch2005,
  title = {Variable {{Neighborhood Search}} for {{Least Squares Clusterwise Regression}}},
  author = {Caporossi, Gilles and Hansen, Pierre},
  date = {2005-01-01},
  journaltitle = {Les Cahiers du GERAD},
  shortjournal = {Les Cahiers du GERAD},
  volume = {2005},
  abstract = {Abstract Clusterwise regression is a technique for clustering data. Instead of using the clas- sical homogeneity or separation criterion, clusterwise regression is based upon the accuracy of a linear regression model associated to each cluster. This model has many advantages, specially for the purpose of data mining, however, the underlying math- ematical model is dicult,to solve due to its large number of local optima. In this paper, we propose the use of the Variable Neighborhood Search metaheuristic (VNS) to improve the quality of the solution. Two perturbation strategies are described and one of them yields a substantial improvement if compared to multistart (the error is reduced by a factor of more than 1.5 on average for the 10 clusters problem). R´esum´e},
  file = {C:\Users\sjelic\Zotero\storage\NZBP6AR6\Caporossi and Hansen - 2005 - Variable Neighborhood Search for Least Squares Clu.pdf}
}

@article{cehEstimatingPerformanceRandom2018,
  title = {Estimating the {{Performance}} of {{Random Forest}} versus {{Multiple Regression}} for {{Predicting Prices}} of the {{Apartments}}},
  author = {Čeh, Marjan and Kilibarda, Milan and Lisec, Anka and Bajat, Branislav},
  date = {2018-05-02},
  journaltitle = {ISPRS International Journal of Geo-Information},
  shortjournal = {IJGI},
  volume = {7},
  number = {5},
  pages = {168},
  issn = {2220-9964},
  doi = {10.3390/ijgi7050168},
  url = {http://www.mdpi.com/2220-9964/7/5/168},
  abstract = {The goal of this study is to analyse the predictive performance of the random forest machine learning technique in comparison to commonly used hedonic models based on multiple regression for the prediction of apartment prices. A data set that includes 7407 records of apartment transactions referring to real estate sales from 2008–2013 in the city of Ljubljana, the capital of Slovenia, was used in order to test and compare the predictive performances of both models. Apparent challenges faced during modelling included (1) the non-linear nature of the prediction assignment task; (2) input data being based on transactions occurring over a period of great price changes in Ljubljana whereby a 28\% decline was noted in six consecutive testing years; and (3) the complex urban form of the case study area. Available explanatory variables, organised as a Geographic Information Systems (GIS) ready dataset, including the structural and age characteristics of the apartments as well as environmental and neighbourhood information were considered in the modelling procedure. All performance measures (R2 values, sales ratios, mean average percentage error (MAPE), coefficient of dispersion (COD)) revealed significantly better results for predictions obtained by the random forest method, which confirms the prospective of this machine learning technique on apartment price prediction.},
  langid = {english},
  file = {C:\Users\sjelic\Zotero\storage\DW4RRMWQ\Čeh et al. - 2018 - Estimating the Performance of Random Forest versus.pdf}
}

@online{chenDeepKrigingSpatiallyDependent2022,
  title = {{{DeepKriging}}: {{Spatially Dependent Deep Neural Networks}} for {{Spatial Prediction}}},
  shorttitle = {{{DeepKriging}}},
  author = {Chen, Wanfang and Li, Yuxiao and Reich, Brian J. and Sun, Ying},
  date = {2022-05-23},
  eprint = {2007.11972},
  eprinttype = {arXiv},
  eprintclass = {cs, stat},
  doi = {10.48550/arXiv.2007.11972},
  url = {http://arxiv.org/abs/2007.11972},
  abstract = {In spatial statistics, a common objective is to predict values of a spatial process at unobserved locations by exploiting spatial dependence. Kriging provides the best linear unbiased predictor using covariance functions and is often associated with Gaussian processes. However, when considering non-linear prediction for non-Gaussian and categorical data, the Kriging prediction is no longer optimal, and the associated variance is often overly optimistic. Although deep neural networks (DNNs) are widely used for general classification and prediction, they have not been studied thoroughly for data with spatial dependence. In this work, we propose a novel DNN structure for spatial prediction, where the spatial dependence is captured by adding an embedding layer of spatial coordinates with basis functions. We show in theory and simulation studies that the proposed DeepKriging method has a direct link to Kriging in the Gaussian case, and it has multiple advantages over Kriging for non-Gaussian and non-stationary data, i.e., it provides non-linear predictions and thus has smaller approximation errors, it does not require operations on covariance matrices and thus is scalable for large datasets, and with sufficiently many hidden neurons, it provides the optimal prediction in terms of model capacity. We further explore the possibility of quantifying prediction uncertainties based on density prediction without assuming any data distribution. Finally, we apply the method to predicting PM2.5 concentrations across the continental United States.},
  pubstate = {prepublished},
  keywords = {Computer Science - Machine Learning,Statistics - Applications,Statistics - Machine Learning,Statistics - Methodology},
  file = {C\:\\Users\\sjelic\\Zotero\\storage\\YII7ITXA\\Chen et al. - 2022 - DeepKriging Spatially Dependent Deep Neural Netwo.pdf;C\:\\Users\\sjelic\\Zotero\\storage\\WZP2E522\\2007.html}
}

@inproceedings{chenSurvivableNetworkDesign2022,
  title = {Survivable {{Network Design Revisited}}: {{Group-Connectivity}}},
  shorttitle = {Survivable {{Network Design Revisited}}},
  booktitle = {2022 {{IEEE}} 63rd {{Annual Symposium}} on {{Foundations}} of {{Computer Science}} ({{FOCS}})},
  author = {Chen, Qingyun and Laekhanukit, Bundit and Liao, Chao and Zhang, Yuhao},
  date = {2022-10},
  pages = {278--289},
  issn = {2575-8454},
  doi = {10.1109/FOCS54457.2022.00033},
  url = {https://ieeexplore.ieee.org/document/9996941},
  abstract = {In the classical survivable network design problem (SNDP), we are given an undirected graph G-(V,E) with costs on edges and a connectivity requirement k(5,t) for each pair of vertices. The goal is to find a minimum-cost subgraph H\textbackslash sqsubseteq G such that every pair (s,t) are connected by k(s,t) edge or (openly) vertex disjoint paths, abbreviated as EC-SNDP and VC-SNDP, respectively. The seminal result of Jain [FOCS’98, Combinatorica’01] gives a 2-approximation algorithm for EC-SNDP, and a decade later, an O(k\textasciicircum 3łog n)- approximation algorithm for VC-SNDP, where k is the largest connectivity requirement, was discovered by Chuzhoy and Khanna [FOCS’09, Theory Comput’12]. While there is a rich literature on point-to-point settings of SNDP, the viable case of connectivity between subsets is still relatively poorly understood. This paper concerns the generalization of SNDP into the subset-to-subset setting, namely Group EC-SNDR We develop the framework, which yields the first non-trivial (true) approximation algorithm for Group. EC-SNDE Previously only a bicriteria approximation algorithm is known for Group EC-SNDP [Chalermsook, Grandoni, and Laekhanukit, SODA’15l, and a true approximation algorithm is known only for the single-source variant with connectivity requirement k(S,T)ın\{0,1,2\} [Gupta, Krishnaswamy, and Ravi, SODA’10; Khandekar, Kortsarz, and Nutov, FSTTCS’09 and Theor Comput. Sci’12].},
  eventtitle = {2022 {{IEEE}} 63rd {{Annual Symposium}} on {{Foundations}} of {{Computer Science}} ({{FOCS}})},
  keywords = {approximation algorithms,Approximation algorithms,Computer science,Costs,network design},
  file = {C\:\\Users\\sjelic\\Zotero\\storage\\88Z3V897\\Chen et al. - 2022 - Survivable Network Design Revisited Group-Connect.pdf;C\:\\Users\\sjelic\\Zotero\\storage\\H6M2FY37\\9996941.html}
}

@article{coffmanEmpiricalPerformanceEvaluation2014,
  title = {An {{Empirical Performance Evaluation}} of {{Relational Keyword Search Techniques}}},
  author = {Coffman, Joel and Weaver, Alfred C.},
  date = {2014-01},
  journaltitle = {IEEE Transactions on Knowledge and Data Engineering},
  volume = {26},
  number = {1},
  pages = {30--42},
  issn = {1558-2191},
  doi = {10.1109/TKDE.2012.228},
  url = {https://ieeexplore.ieee.org/document/6361392},
  abstract = {Extending the keyword search paradigm to relational data has been an active area of research within the database and IR community during the past decade. Many approaches have been proposed, but despite numerous publications, there remains a severe lack of standardization for the evaluation of proposed search techniques. Lack of standardization has resulted in contradictory results from different evaluations, and the numerous discrepancies muddle what advantages are proffered by different approaches. In this paper, we present the most extensive empirical performance evaluation of relational keyword search techniques to appear to date in the literature. Our results indicate that many existing search techniques do not provide acceptable performance for realistic retrieval tasks. In particular, memory consumption precludes many search techniques from scaling beyond small data sets with tens of thousands of vertices. We also explore the relationship between execution time and factors varied in previous evaluations; our analysis indicates that most of these factors have relatively little impact on performance. In summary, our work confirms previous claims regarding the unacceptable performance of these search techniques and underscores the need for standardization in evaluations–standardization exemplified by the IR community.},
  eventtitle = {{{IEEE Transactions}} on {{Knowledge}} and {{Data Engineering}}},
  keywords = {Benchmark testing,Databases,Electronic publishing,empirical evaluation,Encyclopedias,information retrieval,Internet,Keyword search,relational database},
  file = {C:\Users\sjelic\Zotero\storage\AWKPDDND\6361392.html}
}

@article{cordoneExactAlgorithmNode2006,
  title = {An Exact Algorithm for the Node Weighted {{Steiner}} Tree Problem},
  author = {Cordone, Roberto and Trubian, Marco},
  date = {2006-06-01},
  journaltitle = {4OR},
  shortjournal = {4OR},
  volume = {4},
  number = {2},
  pages = {124--144},
  issn = {1614-2411},
  doi = {10.1007/s10288-005-0081-y},
  url = {https://doi.org/10.1007/s10288-005-0081-y},
  abstract = {The Node Weighted Steiner Tree Problem (NW-STP) is a generalization of the Steiner Tree Problem. A lagrangean heuristic presented in EngevallS: StrLBN: 98, and based on the work in Lucena: 92, solves the problem by relaxing an exponential family of generalized subtour elimination constraints and taking into account only the violated ones as the computation proceeds. In EngevallS: StrLBN: 98 the computational results refer to complete graphs up to one hundred vertices. In this paper, we present a branch-and-bound algorithm based on this formulation. Its performance on the instances from the literature confirms the effectiveness of the approach. The experimentation on a newly generated set of benchmark problems, more similar to the real-world applications, shows that the approach is still valid, provided that suitable refinements on the bounding procedures and a preprocessing phase are introduced. The algorithm solves to optimality all of the considered instances up to one thousand vertices, with the exception of 11 hard instances, derived from the literature of a similar problem, the Prize Collecting Steiner Tree Problem.},
  langid = {english},
  keywords = {Prize collecting,Relax-and-cut,Steiner problem},
  file = {C:\Users\sjelic\Zotero\storage\PHNS4YSF\Cordone and Trubian - 2006 - An exact algorithm for the node weighted Steiner t.pdf}
}

@inproceedings{coricGeneticAlgorithmGroup2018,
  title = {A Genetic Algorithm for {{Group Steiner Tree Problem}}},
  booktitle = {2018 41st {{International Convention}} on {{Information}} and {{Communication Technology}}, {{Electronics}} and {{Microelectronics}} ({{MIPRO}})},
  author = {Čorić, Rebeka and Đumić, Mateja and Jelić, Slobodan},
  date = {2018-05},
  pages = {0944--0949},
  doi = {10.23919/MIPRO.2018.8400173},
  url = {https://ieeexplore.ieee.org/document/8400173},
  abstract = {In Group Steiner Tree Problem (GST) we are given a weighted undirected graph and family of subsets of vertices which are called groups. Our objective is to find a minimum-weight subgraph which contains at least one vertex from each group (groups do not have to be disjoint). GST is NP-hard combinatorial optimization problem that arises from many complex real-life problems such as finding substrate-reaction pathways in protein networks, progressive keyword search in relational databases, team formation in social networks, etc. Heuristic methods are extremely important for finding the good-enough solutions in short time. In this paper we present genetic algorithm for solving GST. We also give results of computational experiments with comparisons to optimal solutions.},
  eventtitle = {2018 41st {{International Convention}} on {{Information}} and {{Communication Technology}}, {{Electronics}} and {{Microelectronics}} ({{MIPRO}})},
  keywords = {Approximation algorithms,genetic algorithm,Genetic algorithms,group Steiner tree problem,integer linear programming,Integer linear programming,minimum spanning tree,Optimization,Sociology,Statistics,Steiner trees},
  file = {C:\Users\sjelic\Zotero\storage\4V3IHY3P\8400173.html}
}

@article{czajkowskiSteeringInterpretabilityDecision2023,
  title = {Steering the Interpretability of Decision Trees Using Lasso Regression - an Evolutionary Perspective},
  author = {Czajkowski, Marcin and Jurczuk, Krzysztof and Kretowski, Marek},
  date = {2023-08-01},
  journaltitle = {Information Sciences},
  shortjournal = {Information Sciences},
  volume = {638},
  pages = {118944},
  issn = {0020-0255},
  doi = {10.1016/j.ins.2023.118944},
  url = {https://www.sciencedirect.com/science/article/pii/S0020025523005133},
  abstract = {Since machine and deep learning have made accurate solutions possible, the search for explainable predictors has begun. Decision trees are competitive in tasks that require transparency, but have been underestimated due to their insufficient prediction performance, often caused by generalization issues. It is especially noticeable in the case of model trees, designed to solve regression tasks. Evolutionary tree induction can to some extent counteract this over and under-fitting problem with its global approach. In this paper, we examine whether integrating the lasso estimator in the tree induction process, can help to control the interpretability of the decision tree and/or improve its overall performance. We propose a new evolutionary model tree inducer called Global Lasso Tree. Its novelty lies in regularization of linear models coefficients, in the leaves during the evolutionary search. To reduce the tree's tendency to misfit, a weighted fitness function is used to dynamically balance the trade-off between conflicting objectives which is the tree error and overall complexity. The proposed method was validated on 26 publicly available regression data sets. The empirical study showed that by using the lasso-based regularization technique, we were able to steer the tree's interpretability and thus generate simpler and significantly more accurate trees.},
  keywords = {Evolutionary algorithms,Lasso,Linear regression,Machine learning,Model tree,Regression tree},
  file = {C\:\\Users\\sjelic\\Zotero\\storage\\SRJH2BBM\\Czajkowski et al. - 2023 - Steering the interpretability of decision trees us.pdf;C\:\\Users\\sjelic\\Zotero\\storage\\NLYDSC3T\\S0020025523005133.html}
}

@online{dangeloApproximationAlgorithmsNodeweighted2022,
  title = {Approximation Algorithms for {{Node-weighted Steiner Problems}}: {{Digraphs}} with {{Additive Prizes}} and {{Graphs}} with {{Submodular Prizes}}},
  shorttitle = {Approximation Algorithms for {{Node-weighted Steiner Problems}}},
  author = {D'Angelo, Gianlorenzo and Delfaraz, Esmaeil},
  date = {2022-11-12},
  eprint = {2211.03653},
  eprinttype = {arXiv},
  eprintclass = {cs},
  url = {http://arxiv.org/abs/2211.03653},
  abstract = {In the \textbackslash emph\{budgeted rooted node-weighted Steiner tree\} problem, we are given a graph \$G\$ with \$n\$ nodes, a predefined node \$r\$, two weights associated to each node modelling costs and prizes. The aim is to find a tree in \$G\$ rooted at \$r\$ such that the total cost of its nodes is at most a given budget \$B\$ and the total prize is maximized. In the \textbackslash emph\{quota rooted node-weighted Steiner tree\} problem, we are given a real-valued quota \$Q\$, instead of the budget, and we aim at minimizing the cost of a tree rooted at \$r\$ whose overall prize is at least \$Q\$. For the case of directed graphs with additive prize function, we develop a technique resorting on a standard flow-based linear programming relaxation to compute a tree with good trade-off between prize and cost, which allows us to provide very simple polynomial time approximation algorithms for both the budgeted and the quota problems. For the \textbackslash emph\{budgeted\} problem, our algorithm achieves a bicriteria \$(1+\textbackslash epsilon, O(\textbackslash frac\{1\}\{\textbackslash epsilon\textasciicircum 2\}n\textasciicircum\{2/3\}\textbackslash ln\{n\}))\$-approximation, for any \$\textbackslash epsilon \textbackslash in (0, 1]\$. For the \textbackslash emph\{quota\} problem, our algorithm guarantees a bicriteria approximation factor of \$(2, O(n\textasciicircum\{2/3\}\textbackslash ln\{n\}))\$. Next, by using the flow-based LP, we provide a surprisingly simple polynomial time \$O((1+\textbackslash epsilon)\textbackslash sqrt\{n\} \textbackslash ln \{n\})\$-approximation algorithm for the node-weighted version of the directed Steiner tree problem, for any \$\textbackslash epsilon{$>$}0\$. For the case of undirected graphs with monotone submodular prize functions over subsets of nodes, we provide a polynomial time \$O(\textbackslash frac\{1\}\{\textbackslash epsilon\textasciicircum 3\}\textbackslash sqrt\{n\}\textbackslash log\{n\})\$-approximation algorithm for the budgeted problem that violates the budget constraint by a factor of at most \$1+\textbackslash epsilon\$, for any \$\textbackslash epsilon \textbackslash in (0, 1]\$. Our technique allows us to provide a good approximation also for the quota problem.},
  pubstate = {prepublished},
  keywords = {Computer Science - Computational Complexity,Computer Science - Data Structures and Algorithms},
  file = {C\:\\Users\\sjelic\\Zotero\\storage\\JRHGCA26\\D'Angelo and Delfaraz - 2022 - Approximation algorithms for Node-weighted Steiner.pdf;C\:\\Users\\sjelic\\Zotero\\storage\\N3ANLX5L\\2211.html}
}

@article{dasilvaWeightedClusterwiseLinear2021,
  title = {Weighted {{Clusterwise Linear Regression}} Based on Adaptive Quadratic Form Distance},
  author = {family=Silva, given=Ricardo A. M., prefix=da, useprefix=true and family=Carvalho, given=Francisco de A. T., prefix=de, useprefix=true},
  date = {2021-12-15},
  journaltitle = {Expert Systems with Applications},
  shortjournal = {Expert Systems with Applications},
  volume = {185},
  pages = {115609},
  issn = {0957-4174},
  doi = {10.1016/j.eswa.2021.115609},
  url = {https://www.sciencedirect.com/science/article/pii/S095741742101006X},
  abstract = {The standard approach to Clusterwise Regression is the Clusterwise Linear Regression method. This approach can lead to data over-fitting, and it is not able to distinguish linear relationships in groups of observations well separated in the space of explanatory variables. This paper presents a Weighted Clusterwise Linear Regression method to obtain homogeneous clusters of observations while maintaining a proper fitting for the response variable, by the minimization of an optimization criterion that combines a k-means-like criterion (based on an adaptive quadratic form dissimilarity) in x-space and the criterion of minimum squared residuals of Regression Analysis. The adaptive metric allows automatic weighing or take into account the correlation between explanatory variables under multiple constraints types. We explore six constraints types. Experiments with synthetic and benchmark datasets corroborate the usefulness of the proposed method.},
  keywords = {Adaptive distances,Clustering,Clusterwise regression,Quadratic form distance},
  file = {C\:\\Users\\sjelic\\Zotero\\storage\\V89BSAIJ\\da Silva and de Carvalho - 2021 - Weighted Clusterwise Linear Regression based on ad.pdf;C\:\\Users\\sjelic\\Zotero\\storage\\FCKCABPS\\S095741742101006X.html}
}

@inproceedings{davidovicMatheuristicsBasedVariable2020,
  title = {Matheuristics {{Based}} on {{Variable Neighborhood Search}}},
  author = {Davidović, Tatjana and Matijević, Luka},
  date = {2020-09-01},
  abstract = {Matheuristics represent heuristic optimization methods based on hybridizing exact solvers with metaheuristics. Usually, the exact solvers work on Mathematical Programming (more precisely Mixed Integer Linear Programming, MILP) formulation of the considered problem. Metaheuristic principles are used to define subproblems (e.g., by fixing values for a subset of variables) and exact solver is then invoked to determine values for the remaining variables. The main goal of this paper is to promote three matheuristics that explore Variable Neighborhood Search (VNS) as metaheuristic part. These are Variable Neighborhood Branching (VNB),},
  file = {C:\Users\sjelic\Zotero\storage\66C32HS4\Davidović and Matijević - 2020 - Matheuristics Based on Variable Neighborhood Search.pdf}
}

@inproceedings{davidovicVNSBasedMatheuristicApproach2024,
  title = {{{VNS-Based Matheuristic Approach}} to~{{Group Steiner Tree}} with~{{Problem-Specific Node Release Strategy}}},
  booktitle = {Metaheuristics},
  author = {Davidović, Tatjana and Jelić, Slobodan},
  editor = {Sevaux, Marc and Olteanu, Alexandru-Liviu and Pardo, Eduardo G. and Sifaleras, Angelo and Makboul, Salma},
  date = {2024},
  pages = {344--358},
  publisher = {Springer Nature Switzerland},
  location = {Cham},
  doi = {10.1007/978-3-031-62912-9_32},
  abstract = {For a given undirected graph \$\$G = (V, E)\$\$G=(V,E)with a non-negative weight function \$\$w : E \textbackslash rightarrow \textbackslash mathbb \{R\}\_\{+\}\$\$w:E→R+and subsets \$\$G\_1, \textbackslash dots , G\_k\$\$G1,⋯,Gkof V, the Group Steiner Tree (GST) problem consists of constructing a tree \$\$T = (V\_T, E\_T)\$\$T=(VT,ET)with minimal cost, where \$\$V\_T \textbackslash subseteq V\$\$VT⊆V, \$\$E\_T \textbackslash subseteq E\$\$ET⊆E, and T spans at least one node from each of the groups. We develop a VNS-based metaheuristics approach for solving the GST problem. Our main contribution is that we propose a new problem-specific node release strategy that mimics the steps of a VNS-based heuristic. Instead of exploring different neighborhoods by combinatorially enumerating neighboring solutions, as in classical local search, we use a provably good Integer Linear Programming (ILP) formulation to solve a sequence of subproblems of the original problem. Our approach leads to an improvement over the state-of-the-art Gurobi solver both in terms of quality and runtime of the instances available in the literature.},
  isbn = {978-3-031-62912-9},
  langid = {english},
  file = {C:\Users\sjelic\Zotero\storage\A7Q3P9HB\Davidović and Jelić - 2024 - VNS-Based Matheuristic Approach to Group Steiner T.pdf}
}

@article{demsVarStatisticalComparisonsClassifiers,
  title = {Statistical {{Comparisons}} of {{Classifiers}} over {{Multiple Data Sets}}},
  author = {Demsˇar, Janez and Demsar, Janez},
  abstract = {While methods for comparing two learning algorithms on a single data set have been scrutinized for quite some time already, the issue of statistical tests for comparisons of more algorithms on multiple data sets, which is even more essential to typical machine learning studies, has been all but ignored. This article reviews the current practice and then theoretically and empirically examines several suitable tests. Based on that, we recommend a set of simple, yet safe and robust non-parametric tests for statistical comparisons of classifiers: the Wilcoxon signed ranks test for comparison of two classifiers and the Friedman test with the corresponding post-hoc tests for comparison of more classifiers over multiple data sets. Results of the latter can also be neatly presented with the newly introduced CD (critical difference) diagrams.},
  langid = {english},
  file = {C:\Users\sjelic\Zotero\storage\J6A3W6FD\Demsˇar и Demsar - Statistical Comparisons of Classifiers over Multip.pdf}
}

@article{desarboMaximumLikelihoodMethodology1988,
  title = {A Maximum Likelihood Methodology for Clusterwise Linear Regression},
  author = {DeSarbo, Wayne S. and Cron, William L.},
  date = {1988-09-01},
  journaltitle = {Journal of Classification},
  shortjournal = {Journal of Classification},
  volume = {5},
  number = {2},
  pages = {249--282},
  issn = {1432-1343},
  doi = {10.1007/BF01897167},
  url = {https://doi.org/10.1007/BF01897167},
  abstract = {This paper presents a conditional mixture, maximum likelihood methodology for performing clusterwise linear regression. This new methodology simultaneously estimates separate regression functions and membership inK clusters or groups. A review of related procedures is discussed with an associated critique. The conditional mixture, maximum likelihood methodology is introduced together with the E-M algorithm utilized for parameter estimation. A Monte Carlo analysis is performed via a fractional factorial design to examine the performance of the procedure. Next, a marketing application is presented concerning the evaluations of trade show performance by senior marketing executives. Finally, other potential applications and directions for future research are identified.},
  langid = {english},
  keywords = {Cluster analysis,E-M algorithm,Marketing trade shows,Maximum likelihood estimation,Multiple regression},
  file = {C:\Users\sjelic\Zotero\storage\AM6GSKMU\DeSarbo и Cron - 1988 - A maximum likelihood methodology for clusterwise l.pdf}
}

@inproceedings{devlinBERTPretrainingDeep2019,
  title = {{{BERT}}: {{Pre-training}} of {{Deep Bidirectional Transformers}} for {{Language Understanding}}},
  shorttitle = {{{BERT}}},
  booktitle = {Proceedings of the 2019 {{Conference}} of the {{North American Chapter}} of the {{Association}} for {{Computational Linguistics}}: {{Human Language Technologies}}, {{Volume}} 1 ({{Long}} and {{Short Papers}})},
  author = {Devlin, Jacob and Chang, Ming-Wei and Lee, Kenton and Toutanova, Kristina},
  date = {2019-06},
  pages = {4171--4186},
  publisher = {Association for Computational Linguistics},
  location = {Minneapolis, Minnesota},
  doi = {10.18653/v1/N19-1423},
  url = {https://aclanthology.org/N19-1423},
  abstract = {We introduce a new language representation model called BERT, which stands for Bidirectional Encoder Representations from Transformers. Unlike recent language representation models (Peters et al., 2018a; Radford et al., 2018), BERT is designed to pre-train deep bidirectional representations from unlabeled text by jointly conditioning on both left and right context in all layers. As a result, the pre-trained BERT model can be fine-tuned with just one additional output layer to create state-of-the-art models for a wide range of tasks, such as question answering and language inference, without substantial task-specific architecture modifications. BERT is conceptually simple and empirically powerful. It obtains new state-of-the-art results on eleven natural language processing tasks, including pushing the GLUE score to 80.5 (7.7 point absolute improvement), MultiNLI accuracy to 86.7\% (4.6\% absolute improvement), SQuAD v1.1 question answering Test F1 to 93.2 (1.5 point absolute improvement) and SQuAD v2.0 Test F1 to 83.1 (5.1 point absolute improvement).},
  eventtitle = {{{NAACL-HLT}} 2019},
  file = {C:\Users\sjelic\Zotero\storage\KEESR2GA\Devlin et al. - 2019 - BERT Pre-training of Deep Bidirectional Transform.pdf}
}

@online{dimariLASSOPenalizedClusterwiseLinear2021,
  type = {SSRN Scholarly Paper},
  title = {{{LASSO-Penalized Clusterwise Linear Regression Modeling With Local Least Angle Regression}} ({{L-LARS}})},
  author = {Di Mari, Roberto and Rocci, Roberto and Gattone, Stefano Antonio},
  date = {2021-04-21},
  number = {3832769},
  location = {Rochester, NY},
  doi = {10.2139/ssrn.3832769},
  url = {https://papers.ssrn.com/abstract=3832769},
  abstract = {In clusterwise regression analysis, the goal is to predict a response variable based on a set of explanatory variables, each with cluster-specific effects. Nowadays, the number of candidates is typically large: whereas some of these variables might be useful, some others might contribute very little to the prediction. A well known method to perform variable selection is the LASSO, with calibration done by minimizing the Bayesian Information Criterion (BIC). However, current LASSO-penalized estimators have several disadvantages: only certain types of penalties are considered; the computations might involve approximate schemes and can be very time consuming, with overly complex tuning of the penalty term. This is usually due to the possibly large number of times the estimator must be evaluated for each plausible value of the tuning parameter(s). To ease such computation, we introduce an Expectation Maximization (EM) algorithm with closed-form updates working with a very general version of the LASSO penalty. Such an EM is based on an iterative scheme where the component specific LASSO regression coefficients are computed according to a coordinate descent update. Least Angle Regression is then used to perform covariate selection by evaluating the estimator only once. The advantages of our proposal, in terms of computation time reduction and accuracy of model estimation and selection, are shown by means of a simulation study and illustrated with a real data application.},
  langid = {english},
  pubstate = {prepublished},
  keywords = {Clusterwise Linear Regression,Covariate Selection,Penalized Likelihood,Regularized ML},
  file = {C:\Users\sjelic\Zotero\storage\MXVIFT6S\Di Mari и сар. - 2021 - LASSO-Penalized Clusterwise Linear Regression Mode.pdf}
}

@inproceedings{dingFindingTopkMinCost2007,
  title = {Finding {{Top-k Min-Cost Connected Trees}} in {{Databases}}},
  booktitle = {2007 {{IEEE}} 23rd {{International Conference}} on {{Data Engineering}}},
  author = {Ding, Bolin and Xu Yu, Jeffrey and Wang, Shan and Qin, Lu and Zhang, Xiao and Lin, Xuemin},
  date = {2007-04},
  pages = {836--845},
  issn = {2375-026X},
  doi = {10.1109/ICDE.2007.367929},
  url = {https://ieeexplore.ieee.org/document/4221732},
  abstract = {It is widely realized that the integration of database and information retrieval techniques will provide users with a wide range of high quality services. In this paper, we study processing an l-keyword query, p1, p2, ···, pl, against a relational database which can be modeled as a weighted graph, G(V, E). Here V is a set of nodes (tuples) and E is a set of edges representing foreign key references between tuples. Let Vi V be a set of nodes that contain the keyword pi. We study finding top-k minimum cost connected trees that contain at least one node in every subset Vi, and denote our problem as GST-k. When k = 1, it is known as a minimum cost group Steiner tree problem which is NP-Complete. We observe that the number of keywords, l, is small, and propose a novel parameterized solution, with l as a parameter, to find the optimal GST-1, in time complexity O(3ln + 2l((l + log n)n + m)), where n and m are the numbers of nodes and edges in graph G. Our solution can handle graphs with a large number of nodes. Our GST-1 solution can be easily extended to support GST-k, which outperforms the existing GST-k solutions over both weighted undirected/directed graphs. We conducted extensive experimental studies, and report our finding.},
  eventtitle = {2007 {{IEEE}} 23rd {{International Conference}} on {{Data Engineering}}},
  keywords = {Australia,Costs,Data engineering,Information retrieval,Knowledge engineering,Laboratories,Relational databases,Tree graphs},
  file = {C\:\\Users\\sjelic\\Zotero\\storage\\YBZXBFPD\\Ding et al. - 2007 - Finding Top-k Min-Cost Connected Trees in Database.pdf;C\:\\Users\\sjelic\\Zotero\\storage\\JCDP9GB2\\4221732.html}
}

@article{dreyfusSteinerProblemGraphs1971,
  title = {The Steiner Problem in Graphs},
  author = {Dreyfus, S. E. and Wagner, R. A.},
  date = {1971},
  journaltitle = {Networks},
  volume = {1},
  number = {3},
  pages = {195--207},
  issn = {1097-0037},
  doi = {10.1002/net.3230010302},
  url = {https://onlinelibrary.wiley.com/doi/abs/10.1002/net.3230010302},
  abstract = {An algorithm for solving the Steiner problem on a finite undirected graph is presented. This algorithm computes the set of graph arcs of minimum total length needed to connect a specified set of k graph nodes. If the entire graph contains n nodes, the algorithm requires time proportional to n3/2 + n2 (2k-1 - k - 1) + n(3k-1 - 2k + 3)/2. The time requirement above includes the term n3/2, which can be eliminated if the set of shortest paths connecting each pair of nodes in the graph is available.},
  langid = {english},
  file = {C:\Users\sjelic\Zotero\storage\E8VF9685\net.html}
}

@article{drorGeneralizedSpanningTrees2000,
  title = {Generalized Spanning Trees},
  author = {Dror, M. and Haouari, M. and Chaouachi, J.},
  date = {2000-02-01},
  journaltitle = {European Journal of Operational Research},
  shortjournal = {European Journal of Operational Research},
  volume = {120},
  number = {3},
  pages = {583--592},
  issn = {0377-2217},
  doi = {10.1016/S0377-2217(99)00006-5},
  url = {https://www.sciencedirect.com/science/article/pii/S0377221799000065},
  abstract = {In this paper, we propose a definition for the Generalized Minimal Spanning Tree (GMST) of a graph. The GMST requires spanning at least one node out of every set of disjoint nodes (node partition) in a graph. The analysis of the GMST problem is motivated by real life agricultural settings related to construction of irrigation networks in desert environments. We prove that the GMST problem is NP-hard, and examine a number of heuristic solutions for this problem. Computational experiments comparing these heuristics are presented.},
  keywords = {Agriculture,Genetic algorithms,Irrigation network,Minimum spanning tree,Steiner problem,Traveling salesman problem,Worst case analysis},
  file = {C\:\\Users\\sjelic\\Zotero\\storage\\FBC8PBRA\\Dror et al. - 2000 - Generalized spanning trees.pdf;C\:\\Users\\sjelic\\Zotero\\storage\\SYLBTQE4\\S0377221799000065.html}
}

@inproceedings{druckerImprovingRegressorsUsing1997,
  title = {Improving {{Regressors}} Using {{Boosting Techniques}}},
  booktitle = {Proceedings of the {{Fourteenth International Conference}} on {{Machine Learning}}},
  author = {Drucker, Harris},
  date = {1997-07-08},
  series = {{{ICML}} '97},
  pages = {107--115},
  publisher = {Morgan Kaufmann Publishers Inc.},
  location = {San Francisco, CA, USA},
  isbn = {978-1-55860-486-5},
  file = {C:\Users\sjelic\Zotero\storage\L6B9F4DS\Drucker - 1997 - Improving Regressors using Boosting Techniques.pdf}
}

@article{duinSolvingGroupSteiner2004a,
  title = {Solving Group {{Steiner}} Problems as {{Steiner}} Problems},
  author = {Duin, C. W and Volgenant, A and Voß, S},
  date = {2004-04-01},
  journaltitle = {European Journal of Operational Research},
  shortjournal = {European Journal of Operational Research},
  volume = {154},
  number = {1},
  pages = {323--329},
  issn = {0377-2217},
  doi = {10.1016/S0377-2217(02)00707-5},
  url = {https://www.sciencedirect.com/science/article/pii/S0377221702007075},
  abstract = {The generalized spanning tree or group Steiner problem (GSP) is a generalization of the Steiner problem in graphs (SPG): one requires a tree spanning (at least) one vertex of each subset, given in a family of vertex subsets, while minimizing the sum of the corresponding edge costs. Specialized solution procedures have been developed for this problem. In this paper we investigate the performance of a known but so far neglected transformation to the undirected Steiner problem in graphs. When combined with a recent metaheuristic for the SPG this straightforward approach compares favorably with specialized GSP heuristics. Thus we set a standard for future algorithms.},
  keywords = {Combinatorial optimisation,Generalized minimal spanning tree,Group Steiner problem,Metaheuristics,Pilot method},
  file = {C\:\\Users\\sjelic\\Zotero\\storage\\FDMVCTMW\\Duin et al. - 2004 - Solving group Steiner problems as Steiner problems.pdf;C\:\\Users\\sjelic\\Zotero\\storage\\KGJJV84N\\S0377221702007075.html}
}

@article{efronLeastAngleRegression2004,
  title = {Least Angle Regression},
  author = {Efron, Bradley and Hastie, Trevor and Johnstone, Iain and Tibshirani, Robert},
  date = {2004-04},
  journaltitle = {The Annals of Statistics},
  volume = {32},
  number = {2},
  pages = {407--499},
  publisher = {Institute of Mathematical Statistics},
  issn = {0090-5364, 2168-8966},
  doi = {10.1214/009053604000000067},
  url = {https://projecteuclid.org/journals/annals-of-statistics/volume-32/issue-2/Least-angle-regression/10.1214/009053604000000067.full},
  abstract = {The purpose of model selection algorithms such as All Subsets, Forward Selection and Backward Elimination is to choose a linear model on the basis of the same set of data to which the model will be applied. Typically we have available a large collection of possible covariates from which we hope to select a parsimonious set for the efficient prediction of a response variable. Least Angle Regression (LARS), a new model selection algorithm, is a useful and less greedy version of traditional forward selection methods. Three main properties are derived: (1) A simple modification of the LARS algorithm implements the Lasso, an attractive version of ordinary least squares that constrains the sum of the absolute regression coefficients; the LARS modification calculates all possible Lasso estimates for a given problem, using an order of magnitude less computer time than previous methods. (2) A different LARS modification efficiently implements Forward Stagewise linear regression, another promising new model selection method; this connection explains the similar numerical results previously observed for the Lasso and Stagewise, and helps us understand the properties of both methods, which are seen as constrained versions of the simpler LARS algorithm. (3) A simple approximation for the degrees of freedom of a LARS estimate is available, from which we derive a Cp estimate of prediction error; this allows a principled choice among the range of possible LARS estimates. LARS and its variants are computationally efficient: the paper describes a publicly available algorithm that requires only the same order of magnitude of computational effort as ordinary least squares applied to the full set of covariates.},
  keywords = {62J07,boosting,coefficient paths,Lasso,Linear regression,Variable selection},
  file = {C:\Users\sjelic\Zotero\storage\LZSSJLM6\Efron et al. - 2004 - Least angle regression.pdf}
}

@article{eksiogluVehicleRoutingProblem2009,
  title = {The Vehicle Routing Problem: {{A}} Taxonomic Review},
  shorttitle = {The Vehicle Routing Problem},
  author = {Eksioglu, Burak and Vural, Arif Volkan and Reisman, Arnold},
  date = {2009-11-01},
  journaltitle = {Computers \& Industrial Engineering},
  shortjournal = {Computers \& Industrial Engineering},
  volume = {57},
  number = {4},
  pages = {1472--1483},
  issn = {0360-8352},
  doi = {10.1016/j.cie.2009.05.009},
  url = {https://www.sciencedirect.com/science/article/pii/S0360835209001405},
  abstract = {This paper presents a methodology for classifying the literature of the Vehicle Routing Problem (VRP). VRP as a field of study and practice is defined quite broadly. It is considered to encompass all of the managerial, physical, geographical, and informational considerations as well as the theoretic disciplines impacting this ever emerging-field. Over its lifespan the VRP literature has become quite disjointed and disparate. Keeping track of its development has become difficult because its subject matter transcends several academic disciplines and professions that range from algorithm design to traffic management. Consequently, this paper defines VRP’s domain in its entirety, accomplishes an all-encompassing taxonomy for the VRP literature, and delineates all of VRP’s facets in a parsimonious and discriminating manner. Sample articles chosen for their disparity are classified to illustrate the descriptive power and parsimony of the taxonomy. Moreover, all previously published VRP taxonomies are shown to be relatively myopic; that is, they are subsumed by what is herein presented. Because the VRP literature encompasses esoteric and highly theoretical articles at one extremum and descriptions of actual applications at the other, the article sampling includes the entire range of the VRP literature.},
  keywords = {Classification,Routing,Taxonomy,Vehicle routing,VRP},
  file = {C:\Users\sjelic\Zotero\storage\XAXJI3W8\S0360835209001405.html}
}

@article{faouziTimeSeriesClassification,
  title = {Time {{Series Classification}}: {{A}} Review of {{Algorithms}} and {{Implementations}}},
  author = {Faouzi, Johann},
  journaltitle = {Machine Learning},
  abstract = {Time series classification is a subfield of machine learning with numerous real-life applications. Due to the temporal structure of the input data, standard machine learning algorithms are usually not well suited to work on raw time series. Over the last decades, many algorithms have been proposed to improve the predictive performance and the scalability of state-of-the-art models. Many approaches have been investigated, ranging from deriving new metrics to developing bag-of-words models to imaging time series to artificial neural networks. In this review, we present in detail the major contributions made to this field and mention their most prominent extensions. We dedicate a section to each category of algorithms, with an intuitive introduction on the general approach, detailed theoretical descriptions and explicit illustrations of the major contributions, and mentions of their most prominent extensions. At last, we dedicate a section to publicly available resources, namely data sets and open-source software, for time series classification. A particular emphasis is made on enumerating the availability of the mentioned algorithms in the most popular libraries. The combination of theoretical and practical contents provided in this review will help the readers to easily get started on their own work on time series classification, whether it be theoretical or practical.},
  langid = {english},
  file = {C:\Users\sjelic\Zotero\storage\Q4FAFNXU\Faouzi - Time Series Classification A review of Algorithms.pdf}
}

@inproceedings{farahaniBriefReviewDomain2021,
  title = {A {{Brief Review}} of {{Domain Adaptation}}},
  booktitle = {Advances in {{Data Science}} and {{Information Engineering}}},
  author = {Farahani, Abolfazl and Voghoei, Sahar and Rasheed, Khaled and Arabnia, Hamid R.},
  editor = {Stahlbock, Robert and Weiss, Gary M. and Abou-Nasr, Mahmoud and Yang, Cheng-Ying and Arabnia, Hamid R. and Deligiannidis, Leonidas},
  date = {2021},
  pages = {877--894},
  publisher = {Springer International Publishing},
  location = {Cham},
  abstract = {Classical machine learning assumes that the training and test sets come from the same distributions. Therefore, a model learned from the labeled training data is expected to perform well on the test data. However, this assumption may not always hold in real-world applications where the training and the test data fall from different distributions, due to many factors, e.g., collecting the training and test sets from different sources or having an outdated training set due to the change of data over time. In this case, there would be a discrepancy across domain distributions, and naively applying the trained model on the new dataset may cause degradation in the performance. Domain adaptation is a subfield within machine learning that aims to cope with these types of problems by aligning the disparity between domains such that the trained model can be generalized into the domain of interest. This paper focuses on unsupervised domain adaptation, where the labels are only available in the source domain. It addresses the categorization of domain adaptation from different viewpoints. Besides, it presents some successful shallow and deep domain adaptation approaches that aim to deal with domain adaptation problems.},
  isbn = {978-3-030-71704-9},
  file = {C:\Users\sjelic\Zotero\storage\9CE2VKCG\Farahani et al. - 2020 - A Brief Review of Domain Adaptation.pdf}
}

@article{faustPathwayDiscoveryMetabolic2010,
  title = {Pathway Discovery in Metabolic Networks by Subgraph Extraction},
  author = {Faust, Karoline and Dupont, Pierre and Callut, Jérôme and family=Helden, given=Jacques, prefix=van, useprefix=true},
  date = {2010-05-01},
  journaltitle = {Bioinformatics},
  shortjournal = {Bioinformatics},
  volume = {26},
  number = {9},
  pages = {1211--1218},
  issn = {1367-4803},
  doi = {10.1093/bioinformatics/btq105},
  url = {https://doi.org/10.1093/bioinformatics/btq105},
  abstract = {Motivation: Subgraph extraction is a powerful technique to predict pathways from biological networks and a set of query items (e.g. genes, proteins, compounds, etc.). It can be applied to a variety of different data types, such as gene expression, protein levels, operons or phylogenetic profiles. In this article, we investigate different approaches to extract relevant pathways from metabolic networks. Although these approaches have been adapted to metabolic networks, they are generic enough to be adjusted to other biological networks as well.Results: We comparatively evaluated seven sub-network extraction approaches on 71 known metabolic pathways from Saccharomyces cerevisiae and a metabolic network obtained from MetaCyc. The best performing approach is a novel hybrid strategy, which combines a random walk-based reduction of the graph with a shortest paths-based algorithm, and which recovers the reference pathways with an accuracy of ∼77\%.Availability: Most of the presented algorithms are available as part of the network analysis tool set (NeAT). The kWalks method is released under the GPL3 license.Contact: ~kfaust@ulb.ac.beSupplementary information: ~Supplementary data are available at Bioinformatics online.},
  file = {C\:\\Users\\sjelic\\Zotero\\storage\\9EMSBZVH\\Faust et al. - 2010 - Pathway discovery in metabolic networks by subgrap.pdf;C\:\\Users\\sjelic\\Zotero\\storage\\X9YA3NM5\\199334.html}
}

@article{favarettoAntColonySystem2007,
  title = {Ant Colony System for a {{VRP}} with Multiple Time Windows and Multiple Visits},
  author = {Favaretto, Daniela and Moretti, Elena and Pellegrini, Paola},
  date = {2007-04-01},
  journaltitle = {Journal of Interdisciplinary Mathematics},
  volume = {10},
  number = {2},
  pages = {263--284},
  publisher = {Taylor \& Francis},
  issn = {0972-0502},
  doi = {10.1080/09720502.2007.10700491},
  url = {https://doi.org/10.1080/09720502.2007.10700491},
  abstract = {The Vehicle routing problem with time windows is frequently found in literature, while multiple time windows are not often considered. In this paper a mathematical formulation of the vehicle routing problem with multiple time windows is presented, taking into account periodic constraints. An algorithm based on Ant Colony System is proposed and implemented. Computational results related to a purpose-built benchmark are finally reported.},
  keywords = {ant colony system,Logistics,multiple time windows,vehicle routing problem},
  file = {C:\Users\sjelic\Zotero\storage\YEKZ7YZI\Favaretto et al. - 2007 - Ant colony system for a VRP with multiple time windows and multiple visits.pdf}
}

@article{fawazDeepLearningTime2019,
  title = {Deep Learning for Time Series Classification: A Review},
  shorttitle = {Deep Learning for Time Series Classification},
  author = {Fawaz, Hassan Ismail and Forestier, Germain and Weber, Jonathan and Idoumghar, Lhassane and Muller, Pierre-Alain},
  date = {2019-07-01},
  journaltitle = {Data Mining and Knowledge Discovery},
  shortjournal = {Data Min Knowl Disc},
  volume = {33},
  number = {4},
  eprint = {1809.04356},
  eprinttype = {arXiv},
  eprintclass = {cs, stat},
  pages = {917--963},
  issn = {1384-5810, 1573-756X},
  doi = {10.1007/s10618-019-00619-1},
  url = {http://arxiv.org/abs/1809.04356},
  abstract = {Time Series Classification (TSC) is an important and challenging problem in data mining. With the increase of time series data availability, hundreds of TSC algorithms have been proposed. Among these methods, only a few have considered Deep Neural Networks (DNNs) to perform this task. This is surprising as deep learning has seen very successful applications in the last years. DNNs have indeed revolutionized the field of computer vision especially with the advent of novel deeper architectures such as Residual and Convolutional Neural Networks. Apart from images, sequential data such as text and audio can also be processed with DNNs to reach state-of-the-art performance for document classification and speech recognition. In this article, we study the current state-of-the-art performance of deep learning algorithms for TSC by presenting an empirical study of the most recent DNN architectures for TSC. We give an overview of the most successful deep learning applications in various time series domains under a unified taxonomy of DNNs for TSC. We also provide an open source deep learning framework to the TSC community where we implemented each of the compared approaches and evaluated them on a univariate TSC benchmark (the UCR/UEA archive) and 12 multivariate time series datasets. By training 8,730 deep learning models on 97 time series datasets, we propose the most exhaustive study of DNNs for TSC to date.},
  keywords = {Computer Science - Artificial Intelligence,Computer Science - Machine Learning,Statistics - Machine Learning},
  file = {C\:\\Users\\sjelic\\Zotero\\storage\\8BAALAVI\\Fawaz et al. - 2019 - Deep learning for time series classification a re.pdf;C\:\\Users\\sjelic\\Zotero\\storage\\J93AIDQN\\1809.html}
}

@article{fengKernelizedElasticNet2016,
  title = {Kernelized Elastic Net Regularization: {{Generalization}} Bounds, and Sparse Recovery},
  shorttitle = {Kernelized Elastic Net Regularization},
  author = {Feng, Yunlong and Lv, Shao-Gao and Hang, Hanyuan and Suykens, Johan A. K.},
  date = {2016-03-01},
  journaltitle = {Neural Comput.},
  volume = {28},
  number = {3},
  pages = {525--562},
  issn = {0899-7667},
  doi = {10.1162/NECO_a_00812},
  url = {https://doi.org/10.1162/NECO_a_00812},
  abstract = {Kernelized elastic net regularization KENReg is a kernelization of the well-known elastic net regularization Zou \& Hastie, 2005. The kernel in KENReg is not required to be a Mercer kernel since it learns from a kernelized dictionary in the coefficient space. Feng, Yang, Zhao, Lv, and Suykens 2014 showed that KENReg has some nice properties including stability, sparseness, and generalization. In this letter, we continue our study on KENReg by conducting a refined learning theory analysis. This letter makes the following three main contributions. First, we present refined error analysis on the generalization performance of KENReg. The main difficulty of analyzing the generalization error of KENReg lies in characterizing the population version of its empirical target function. We overcome this by introducing a weighted Banach space associated with the elastic net regularization. We are then able to conduct elaborated learning theory analysis and obtain fast convergence rates under proper complexity and regularity assumptions. Second, we study the sparse recovery problem in KENReg with fixed design and show that the kernelization may improve the sparse recovery ability compared to the classical elastic net regularization. Finally, we discuss the interplay among different properties of KENReg that include sparseness, stability, and generalization. We show that the stability of KENReg leads to generalization, and its sparseness confidence can be derived from generalization. Moreover, KENReg is stable and can be simultaneously sparse, which makes it attractive theoretically and practically.},
  file = {C:\Users\sjelic\Zotero\storage\VJTYLF9I\Feng et al. - 2016 - Kernelized elastic net regularization Generalization bounds, and sparse recovery.pdf}
}

@article{ferreiraFormulationsGroupSteiner2006,
  title = {Some Formulations for the Group Steiner Tree Problem},
  author = {Ferreira, Carlos E. and family=Oliveira Filho, given=Fernando M., prefix=de, useprefix=true},
  date = {2006-08-15},
  journaltitle = {Discrete Applied Mathematics},
  shortjournal = {Discrete Applied Mathematics},
  series = {Traces of the {{Latin American Conference}} on {{Combinatorics}}, {{Graphs}} and {{Applications}}},
  volume = {154},
  number = {13},
  pages = {1877--1884},
  issn = {0166-218X},
  doi = {10.1016/j.dam.2006.03.028},
  url = {https://www.sciencedirect.com/science/article/pii/S0166218X06001429},
  abstract = {The group Steiner tree problem consists of, given a graph G, a collection R of subsets of V(G) and a cost c(e) for each edge of G, finding a minimum-cost subtree that connects at least one vertex from each R∈R. It is a generalization of the well-known Steiner tree problem that arises naturally in the design of VLSI chips. In this paper, we study a polyhedron associated with this problem and some extended formulations. We give facet defining inequalities and explore the relationship between the group Steiner tree problem and other combinatorial optimization problems.},
  keywords = {Branch-and-cut algorithms,Combinatorial optimization,Integer programming formulations,Polyhedral combinatorics}
}

@article{ferreiraNewReductionTechniques2007,
  title = {New {{Reduction Techniques}} for the {{Group Steiner Tree Problem}}},
  author = {Ferreira, Carlos Eduardo and family=Oliveira Filho, given=Fernando M., prefix=de, useprefix=true},
  date = {2007-01},
  journaltitle = {SIAM Journal on Optimization},
  shortjournal = {SIAM J. Optim.},
  volume = {17},
  number = {4},
  pages = {1176--1188},
  publisher = {{Society for Industrial and Applied Mathematics}},
  issn = {1052-6234},
  doi = {10.1137/040610891},
  url = {https://epubs.siam.org/doi/10.1137/040610891},
  abstract = {The classical Steiner tree problem in weighted graphs seeks a minimum weight connected subgraph containing a given subset of the vertices (terminals). We present a new polynomial-time heuristic that achieves a best-known approximation ratio of 1+ln⁡32≈1.55 for general graphs and best-known approximation ratios of ≈1.28 for both quasi-bipartite graphs (i.e., where no two nonterminals are adjacent) and complete graphs with edge weights 1 and 2. Our method is considerably simpler and easier to implement than previous approaches. We also prove the first known nontrivial performance bound (1.5⋅ OPT) for the iterated 1-Steiner heuristic of Kahng and Robins in quasi-bipartite graphs.}
}

@article{freundDecisionTheoreticGeneralizationOnLine1997,
  title = {A {{Decision-Theoretic Generalization}} of {{On-Line Learning}} and an {{Application}} to {{Boosting}}},
  author = {Freund, Yoav and Schapire, Robert E},
  date = {1997-08-01},
  journaltitle = {Journal of Computer and System Sciences},
  shortjournal = {Journal of Computer and System Sciences},
  volume = {55},
  number = {1},
  pages = {119--139},
  issn = {0022-0000},
  doi = {10.1006/jcss.1997.1504},
  url = {https://www.sciencedirect.com/science/article/pii/S002200009791504X},
  abstract = {In the first part of the paper we consider the problem of dynamically apportioning resources among a set of options in a worst-case on-line framework. The model we study can be interpreted as a broad, abstract extension of the well-studied on-line prediction model to a general decision-theoretic setting. We show that the multiplicative weight-update Littlestone–Warmuth rule can be adapted to this model, yielding bounds that are slightly weaker in some cases, but applicable to a considerably more general class of learning problems. We show how the resulting learning algorithm can be applied to a variety of problems, including gambling, multiple-outcome prediction, repeated games, and prediction of points in Rn. In the second part of the paper we apply the multiplicative weight-update technique to derive a new boosting algorithm. This boosting algorithm does not require any prior knowledge about the performance of the weak learning algorithm. We also study generalizations of the new boosting algorithm to the problem of learning functions whose range, rather than being binary, is an arbitrary finite set or a bounded segment of the real line.},
  file = {C:\Users\sjelic\Zotero\storage\HXVTC27Z\Freund and Schapire - 1997 - A Decision-Theoretic Generalization of On-Line Learning and an Application to Boosting.pdf}
}

@article{friedmanGreedyFunctionApproximation2001,
  title = {Greedy {{Function Approximation}}: {{A Gradient Boosting Machine}}},
  shorttitle = {Greedy {{Function Approximation}}},
  author = {Friedman, Jerome H.},
  date = {2001},
  journaltitle = {The Annals of Statistics},
  volume = {29},
  number = {5},
  eprint = {2699986},
  eprinttype = {jstor},
  pages = {1189--1232},
  publisher = {Institute of Mathematical Statistics},
  issn = {0090-5364},
  url = {https://www.jstor.org/stable/2699986},
  abstract = {Function estimation/approximation is viewed from the perspective of numerical optimization in function space, rather than parameter space. A connection is made between stagewise additive expansions and steepest-descent minimization. A general gradient descent "boosting" paradigm is developed for additive expansions based on any fitting criterion. Specific algorithms are presented for least-squares, least absolute deviation, and Huber-M loss functions for regression, and multiclass logistic likelihood for classification. Special enhancements are derived for the particular case where the individual additive components are regression trees, and tools for interpreting such "TreeBoost" models are presented. Gradient boosting of regression trees produces competitive, highly robust, interpretable procedures for both regression and classification, especially appropriate for mining less than clean data. Connections between this approach and the boosting methods of Freund and Shapire and Friedman, Hastie and Tibshirani are discussed.},
  file = {C:\Users\sjelic\Zotero\storage\MCUEZ95I\Friedman - 2001 - Greedy Function Approximation A Gradient Boosting Machine.pdf}
}

@article{gargPolylogarithmicApproximationAlgorithm2000,
  title = {A {{Polylogarithmic Approximation Algorithm}} for the {{Group Steiner Tree Problem}}},
  author = {Garg, Naveen and Konjevod, Goran and Ravi, R.},
  date = {2000-10-01},
  journaltitle = {Journal of Algorithms},
  shortjournal = {Journal of Algorithms},
  volume = {37},
  number = {1},
  pages = {66--84},
  issn = {0196-6774},
  doi = {10.1006/jagm.2000.1096},
  url = {https://www.sciencedirect.com/science/article/pii/S0196677400910964},
  abstract = {Given a weighted graph with some subsets of vertices called groups, the group Steiner tree problem is to find a minimum-weight subgraph which contains at least one vertex from each group. We give a randomized algorithm with a polylogarithmic approximation guarantee for the group Steiner tree problem. The previous best approximation guarantee was O(i2k1/i) in time O(nik2i) (Charikar, Chekuri, Goel, and Guha). Our algorithm also improves existing approximation results for network design problems with location-based constraints and for the symmetric generalized traveling salesman problem.},
  langid = {english},
  keywords = {approximation algorithms,network design,randomized rounding,set cover,Steiner tree,tree decompositions},
  file = {C:\Users\sjelic\Zotero\storage\7JRJCAH3\Garg et al. - 2000 - A Polylogarithmic Approximation Algorithm for the .pdf}
}

@inproceedings{gehringConvolutionalSequenceSequence2017,
  title = {Convolutional Sequence to Sequence Learning},
  booktitle = {Proceedings of the 34th {{International Conference}} on {{Machine Learning}} - {{Volume}} 70},
  author = {Gehring, Jonas and Auli, Michael and Grangier, David and Yarats, Denis and Dauphin, Yann N.},
  date = {2017-08-06},
  series = {{{ICML}}'17},
  pages = {1243--1252},
  publisher = {JMLR.org},
  location = {Sydney, NSW, Australia},
  abstract = {The prevalent approach to sequence to sequence learning maps an input sequence to a variable length output sequence via recurrent neural networks. We introduce an architecture based entirely on convolutional neural networks. Compared to recurrent models, computations over all elements can be fully parallelized during training to better exploit the GPU hardware and optimization is easier since the number of non-linearities is fixed and independent of the input length. Our use of gated linear units eases gradient propagation and we equip each decoder layer with a separate attention module. We outperform the accuracy of the deep LSTM setup of Wu et al. (2016) on both WMT'14 English-German and WMT'14 English-French translation at an order of magnitude faster speed, both on GPU and CPU.*},
  file = {C:\Users\sjelic\Zotero\storage\XQTGVSXJ\Gehring et al. - 2017 - Convolutional sequence to sequence learning.pdf}
}

@online{gitmanNovelPredictionTechniques2018,
  title = {Novel {{Prediction Techniques Based}} on {{Clusterwise Linear Regression}}},
  author = {Gitman, Igor and Chen, Jieshi and Lei, Eric and Dubrawski, Artur},
  date = {2018-04-28},
  eprint = {1804.10742},
  eprinttype = {arXiv},
  eprintclass = {cs, stat},
  doi = {10.48550/arXiv.1804.10742},
  url = {http://arxiv.org/abs/1804.10742},
  abstract = {In this paper we explore different regression models based on Clusterwise Linear Regression (CLR). CLR aims to find the partition of the data into \$k\$ clusters, such that linear regressions fitted to each of the clusters minimize overall mean squared error on the whole data. The main obstacle preventing to use found regression models for prediction on the unseen test points is the absence of a reasonable way to obtain CLR cluster labels when the values of target variable are unknown. In this paper we propose two novel approaches on how to solve this problem. The first approach, predictive CLR builds a separate classification model to predict test CLR labels. The second approach, constrained CLR utilizes a set of user-specified constraints that enforce certain points to go to the same clusters. Assuming the constraint values are known for the test points, they can be directly used to assign CLR labels. We evaluate these two approaches on three UCI ML datasets as well as on a large corpus of health insurance claims. We show that both of the proposed algorithms significantly improve over the known CLR-based regression methods. Moreover, predictive CLR consistently outperforms linear regression and random forest, and shows comparable performance to support vector regression on UCI ML datasets. The constrained CLR approach achieves the best performance on the health insurance dataset, while enjoying only \$\textbackslash approx 20\$ times increased computational time over linear regression.},
  pubstate = {prepublished},
  keywords = {Computer Science - Machine Learning,Statistics - Machine Learning},
  file = {C\:\\Users\\sjelic\\Zotero\\storage\\N2K6CVW9\\Gitman и сар. - 2018 - Novel Prediction Techniques Based on Clusterwise L.pdf;C\:\\Users\\sjelic\\Zotero\\storage\\HQ6XMVK5\\1804.html}
}

@article{glicksmanApproximationAlgorithmsGroup2008,
  title = {Approximation Algorithms for Group Prize-Collecting and Location-Routing Problems},
  author = {Glicksman, Hagai and Penn, Michal},
  date = {2008-10-28},
  journaltitle = {Discrete Applied Mathematics},
  shortjournal = {Discrete Applied Mathematics},
  series = {Cologne/{{Twente Workshop}} on {{Graphs}} and {{Combinatorial Optimization}}},
  volume = {156},
  number = {17},
  pages = {3238--3247},
  issn = {0166-218X},
  doi = {10.1016/j.dam.2008.05.013},
  url = {https://www.sciencedirect.com/science/article/pii/S0166218X08002254},
  abstract = {In this paper we develop approximation algorithms for generalizations of the following three known combinatorial optimization problems, the Prize-Collecting Steiner Tree problem, the Prize-Collecting Travelling Salesman Problem and a Location-Routing problem. Given a graph G=(V,E) on n vertices and a length function on its edges, in the grouped versions of the above mentioned problems we assume that V is partitioned into k+1 groups, \{V0,V1,…,Vk\}, with a penalty function on the groups. In the Group Prize-Collecting Steiner Tree problem the aim is to find S, a collection of groups of V and a tree spanning the rest of the groups not in S, so as to minimize the sum of the costs of the edges in the tree and the costs of the groups in S. The Group Prize-Collecting Travelling Salesman Problem, is defined analogously. In the Group Location-Routing problem the customer vertices are partitioned into groups and one has to select simultaneously a subset of depots to be opened and a collection of tours that covers the customer groups. The goal is to minimize the costs of the tours plus the fixed costs of the opened depots. We give a (2−1n−1)I-approximation algorithm for each of the three problems, where I is the cardinality of the largest group.},
  langid = {english},
  keywords = {Algorithm,Approximation,Group,Location,Prize-collecting,Travelling salesman},
  file = {C:\Users\sjelic\Zotero\storage\XQ8INJGX\Glicksman and Penn - 2008 - Approximation algorithms for group prize-collectin.pdf}
}

@article{goemansSteinerTreePolytope1994,
  title = {The {{Steiner}} Tree Polytope and Related Polyhedra},
  author = {Goemans, Michel X.},
  date = {1994-01-01},
  journaltitle = {Mathematical Programming},
  shortjournal = {Mathematical Programming},
  volume = {63},
  number = {1},
  pages = {157--182},
  issn = {1436-4646},
  doi = {10.1007/BF01582064},
  url = {https://doi.org/10.1007/BF01582064},
  abstract = {We consider the vertex-weighted version of the undirected Steiner tree problem. In this problem, a cost is incurred both for the vertices and the edges present in the Steiner tree. We completely describe the associated polytope by linear inequalities when the underlying graph is series—parallel. For general graphs, this formulation can be interpreted as a (partial) extended formulation for the Steiner tree problem. By projecting this formulation, we obtain some very large classes of facet-defining valid inequalities for the Steiner tree polytope.},
  langid = {english},
  keywords = {facets,formulations,polyhedral characterization,projection,series—parallel graphs,Steiner tree}
}

@article{greenbergPerformanceSituVs2022,
  title = {Performance of in Situ vs Laboratory Mid-Infrared Soil Spectroscopy Using Local and Regional Calibration Strategies},
  author = {Greenberg, Isabel and Seidel, Michael and Vohland, Michael and Koch, Heinz-Josef and Ludwig, Bernard},
  date = {2022-03-01},
  journaltitle = {Geoderma},
  shortjournal = {Geoderma},
  volume = {409},
  pages = {115614},
  issn = {0016-7061},
  doi = {10.1016/j.geoderma.2021.115614},
  url = {https://www.sciencedirect.com/science/article/pii/S0016706121006947},
  abstract = {Comparison of in situ mid-infrared spectroscopy (MIRS) with laboratory MIRS is required to demonstrate the accuracy of field-scale prediction of soil properties. Application of MIRS to investigate soil management questions must also be tested. Our objectives were therefore to determine i) the accuracy of lab vs in situ calibrations using various numbers of local and/or regional soils for prediction of organic carbon (OC), total nitrogen (TN), clay and pH; ii) effects of soil moisture content and variability on model performance for coarser and finer soils; and iii) if the method of OC determination (dry combustion vs MIRS-estimation) affects evaluation of tillage effects. Surface field MIRS measurements were made at three loess sites in Germany, each featuring three tillage treatments. Material (0–2~cm) was collected for lab MIRS measurements on dried/ground ({$<$}0.2~mm) soil and determination of OC, TN, clay and pH. Spectral Principal Component Analysis (PCA) was conducted and partial least squares regression models were created for several calibration strategies: 1) local calibrations trained with n~=~40 or 20 soils and tested with n~=~110 soils from the same site; 2) regional calibrations trained with n~=~150 or 38 soils from two sites and validated with n~=~110 soils from the third site; 3) regional calibrations trained with n~=~150 or 38 soils from two sites and n~=~20 double- or n~=~10 quadruple-weighted “spiked” soils selected from the spectral PCA to be representative of the third site, and validation with n~=~110 soils also from the third site. Spiking regional calibrations with local soils generally improved accuracy and decreased performance variability, though there were typically diminishing marginal returns to accuracy from increasing the number of local soils. The first two principal components of the lab-MIRS PCA correlated with OC, TN, clay and pH, while the field-MIRS PCA was dominated by soil moisture effects. Lab outperformed field MIRS for all models and properties. Lab MIRS n~=~38 regional models were highly accurate for OC (ratio of prediction to interquartile distance (RPIQ)~=~4.3) and TN (RPIQ~=~6.7), and estimates detected the same significant differences between tillage treatments as analysis conducted with measured values—thus, small regional models can be considered optimal (balancing accuracy and workload). For field MIRS prediction of OC and TN, calibrations with 150 regional or 38 regional plus 10 quadruple-weighted local soils achieved satisfactory accuracy (RPIQ~≥~1.89). Although predicted changes to OC in response to tillage were more biased for field MIRS, agreement with measured effects was achieved with n~=~40 local models or spiked regional models. Thus, the higher efficiency of field measurement is counterbalanced by a more arduous calibration process to achieve satisfactory accuracy. Accuracies for clay (RPIQ~=~0.89–2.8) and pH (RPIQ~=~0.60–3.2) were lower and more variable than OC and TN for both devices—thus, spiking calibrations and using more soils than OC/TN calibrations are recommended. Soil moisture more negatively affected OC prediction than clay prediction. No simple trend was established for the performances of soil subsets with low, high or variable moisture content, but accuracy was most negatively affected by moisture for the site with the highest sand content.},
  keywords = {Field spectroscopy,Handheld spectrometer,MIR,Portable spectrometer,Soil moisture,Soil organic carbon},
  file = {C\:\\Users\\sjelic\\Zotero\\storage\\ME673EY7\\Greenberg et al. - 2022 - Performance of in situ vs laboratory mid-infrared .pdf;C\:\\Users\\sjelic\\Zotero\\storage\\LR5DPKCB\\S0016706121006947.html}
}

@inproceedings{guerrerovazquezRepeatedMeasuresMultiple2001,
  title = {Repeated {{Measures Multiple Comparison Procedures Applied}} to {{Model Selection}} in {{Neural Networks}}},
  booktitle = {Bio-{{Inspired Applications}} of {{Connectionism}}},
  author = {Guerrero Vázquez, Elisa and Yañez Escolano, Andrés and Galindo Riaño, Pedro and Pizarro Junquera, Joaquín},
  editor = {Mira, José and Prieto, Alberto},
  date = {2001},
  series = {Lecture {{Notes}} in {{Computer Science}}},
  pages = {88--95},
  publisher = {Springer},
  location = {Berlin, Heidelberg},
  doi = {10.1007/3-540-45723-2_10},
  abstract = {One of the main research concern in neural networks is to find the appropriate network size in order to minimize the trade-off between overfitting and poor approximation. In this paper the choice among different competing models that fit to the same data set is faced when statistical methods for model comparison are applied. The study has been conducted to find a range of models that can work all the same as the cost of complexity varies. If they do not, then the generalization error estimates should be about the same among the set of models. If they do, then the estimates should be different and our job would consist on analyzing pairwise differences between the least generalization error estimate and each one of the range, in order to bound the set of models which might result in an equal performance. This method is illustrated applied to polynomial regression and RBF neural networks.},
  isbn = {978-3-540-45723-7},
  langid = {english}
}

@inproceedings{guhaEfficientRecoveryPower1999,
  title = {Efficient Recovery from Power Outage (Extended Abstract)},
  booktitle = {Proceedings of the Thirty-First Annual {{ACM}} Symposium on {{Theory}} of {{Computing}}},
  author = {Guha, Sudipto and Moss, Anna and Naor, Joseph (Seffi) and Schieber, Baruch},
  date = {1999-05-01},
  series = {{{STOC}} '99},
  pages = {574--582},
  publisher = {Association for Computing Machinery},
  location = {New York, NY, USA},
  doi = {10.1145/301250.301406},
  url = {https://dl.acm.org/doi/10.1145/301250.301406},
  isbn = {978-1-58113-067-6},
  file = {C:\Users\sjelic\Zotero\storage\QY3EGN9Q\Guha et al. - 1999 - Efficient recovery from power outage (extended abs.pdf}
}

@inproceedings{halperinPolylogarithmicInapproximability2003,
  title = {Polylogarithmic Inapproximability},
  booktitle = {Proceedings of the Thirty-Fifth Annual {{ACM}} Symposium on {{Theory}} of Computing},
  author = {Halperin, Eran and Krauthgamer, Robert},
  date = {2003-06-09},
  series = {{{STOC}} '03},
  pages = {585--594},
  publisher = {Association for Computing Machinery},
  location = {New York, NY, USA},
  doi = {10.1145/780542.780628},
  url = {https://doi.org/10.1145/780542.780628},
  abstract = {We provide the first hardness result of a polylogarithmic approximation ratio for a natural NP-hard optimization problem. We show that for every fixed ε{$>$}0, the GROUP-STEINER-TREE problem admits no efficient log2-ε k approximation, where k denotes the number of groups (or, alternatively, the input size), unless NP has quasi polynomial Las-Vegas algorithms. This hardness result holds even for input graphs which are Hierarchically Well-Separated Trees, introduced by Bartal [FOCS, 1996]. For these trees (and also for general trees), our bound is nearly tight with the log-squared approximation currently known. Our results imply that for every fixed ε{$>$}0, the DIRECTED-STEINER TREE problem admits no log2-ε n--approximation, where n is the number of vertices in the graph, under the same complexity assumption.},
  isbn = {978-1-58113-674-6},
  keywords = {approximation algorithms,hardness of approximation,integrality ratio,polylogarithmic approximation,Steiner tree}
}

@article{hansenVariableNeighborhoodSearch2006,
  title = {Variable Neighborhood Search and Local Branching},
  author = {Hansen, Pierre and Mladenović, Nenad and Urošević, Dragan},
  date = {2006-10},
  journaltitle = {Computers \& Operations Research},
  shortjournal = {Computers \& Operations Research},
  volume = {33},
  number = {10},
  pages = {3034--3045},
  issn = {03050548},
  doi = {10.1016/j.cor.2005.02.033},
  url = {https://linkinghub.elsevier.com/retrieve/pii/S0305054805000894},
  abstract = {In this paper we develop a variable neighborhood search (VNS) heuristic for solving mixed-integer programs (MIPs). It uses CPLEX, the general-purpose MIP solver, as a black-box. Neighborhoods around the incumbent solution are defined by adding constraints to the original problem, as suggested in the recent local branching (LB) method of Fischetti and Lodi (Mathematical Programming Series B 2003;98:23–47). Both LB and VNS use the same tools: CPLEX and the same definition of the neighborhoods around the incumbent. However, our VNS is simpler and more systematic in neighborhood exploration. Consequently, within the same time limit, we were able to improve 14 times the best known solution from the set of 29 hard problem instances used to test LB.},
  langid = {english},
  file = {C:\Users\sjelic\Zotero\storage\NJ9ILBF9\Hansen и сар. - 2006 - Variable neighborhood search and local branching.pdf}
}

@article{haoTransferLearningCrop2020,
  title = {Transfer {{Learning}} for {{Crop}} Classification with {{Cropland Data Layer}} Data ({{CDL}}) as Training Samples},
  author = {Hao, Pengyu and Di, Liping and Zhang, Chen and Guo, Liying},
  date = {2020-09},
  journaltitle = {Science of The Total Environment},
  shortjournal = {Science of The Total Environment},
  volume = {733},
  pages = {138869},
  issn = {00489697},
  doi = {10.1016/j.scitotenv.2020.138869},
  url = {https://linkinghub.elsevier.com/retrieve/pii/S004896972032386X},
  abstract = {Training samples is fundamental for crop mapping from remotely sensed images, but difficult to acquire in many regions through ground survey, causing significant challenge for crop mapping in these regions. In this paper, a transfer learning (TL) workflow is proposed to use the classification model trained in contiguous U.S.A. (CONUS) to identify crop types in other regions. The workflow is based on fact that same crop growing in different regions of world has similar temporal growth pattern. This study selected high confidence pixels across CONUS in the Cropland Data Layer (CDL) and corresponding 30-m 15-day composited NDVI time series generated from harmonized Landat-8 and Sentinel-2 (HLS) data as training samples, trained the Random Forest (RF) classification models and then applied the models to identify crop types in three test regions, namely Hengshui in China (HS), Alberta in Canada (AB), and Nebraska in USA (NE). NDVI time series with different length were used to identify crops, the effect of time-series length on classification accuracies were then evaluated. Furthermore, local training samples in the three test regions were collected and used to identify crops (LO) for comparison. Results showed that overall classification accuracies in HS, AB and NE were 97.79\%, 86.45\% and 94.86\%, respectively, when using TL with NDVI time series of the entire growing season for classification. However, LO could achieve higher classification accuracies earlier than TL. Because the training samples were collected across USA containing multiple growth conditions, it increased the potential that the crop growth environment in test regions could be similar to those of the training samples; but also led to situation that different crops had similar NDVI time series, which caused lower TL classification accuracy in HS at early-season. Generally, this study provides new options for crop classification in regions of training samples shortage.},
  langid = {english},
  file = {C:\Users\sjelic\Zotero\storage\H4EXVBSQ\Hao et al. - 2020 - Transfer Learning for Crop classification with Cro.pdf}
}

@article{hennigModelsMethodsClusterwise1998,
  title = {Models {{And Methods For Clusterwise Linear Regression}}},
  author = {Hennig, Christian},
  date = {1998-12-19},
  issn = {978-3-540-65855-9},
  doi = {10.1007/978-3-642-60187-3_17},
  abstract = {: Three models for linear regression clustering are given, and corresponding methods for classification and parameter estimation are developed and discussed: The mixture model with fixed regressors (ML-estimation), the fixed partition model with fixed regressors (ML-estimation), and the mixture model with random regressors (Fixed Point Clustering). The number of clusters is treated as unknown. The approaches are compared via an application to Fisher's Iris data. By the way, a broadly ignored feature of these data is discovered. 1 Introduction Cluster analysis problems based on stochastic models can be divided into two classes: 1. A cluster is considered as a subset of the data points, which can be modeled adequately by a distribution from a class of cluster reference distributions (c.r.d.). These distributions are chosen to reflect the meaning of homogeneity with respect to the certain data analysis problem. Therefore c.r.d. are often unimodal. If the class of c.r.d. is parametric, th...},
  file = {C:\Users\sjelic\Zotero\storage\DEXW6I5W\Hennig - 1998 - Models And Methods For Clusterwise Linear Regressi.pdf}
}

@article{hongPotentialGloballyDistributed2024,
  title = {Potential of Globally Distributed Topsoil Mid-Infrared Spectral Library for Organic Carbon Estimation},
  author = {Hong, Yongsheng and Sanderman, Jonathan and Hengl, Tomislav and Chen, Songchao and Wang, Nan and Xue, Jie and Zhuo, Zhiqing and Peng, Jie and Li, Shuo and Chen, Yiyun and Liu, Yaolin and Mouazen, Abdul Mounem and Shi, Zhou},
  date = {2024-02-01},
  journaltitle = {CATENA},
  shortjournal = {CATENA},
  volume = {235},
  pages = {107628},
  issn = {0341-8162},
  doi = {10.1016/j.catena.2023.107628},
  url = {https://www.sciencedirect.com/science/article/pii/S0341816223007191},
  abstract = {Accurate monitoring of soil organic carbon (SOC) is critical for sustainable management of soil for improving its quality, function, and carbon sequestration. As a nondestructive, efficient, and low-cost technique, mid-infrared (MIR) spectroscopy has shown a great potential in rapid estimation of SOC, despite limited studies of the global scale. The objective of this work was to use a globally distributed topsoil MIR spectral library with 33,039 samples to predict SOC using different modeling methods. Effects of nine fractional-order derivatives (FODs) on the predicted accuracy of SOC were evaluated using four regression algorithms (i.e., ratio index-based linear regression, RI-LR; partial least squares regression, PLSR; Cubist; convolutional neural network, CNN). Square-root transformation to SOC data was performed to minimize the skewness and non-linearity. Results indicated FOD to capture the subtle spectral details related to SOC, leading to improved predictions that may not be possible by the raw absorbance and common integer-order derivatives. Concerning the RI-LR models, the optimal validation result for SOC was obtained by 0.75-order derivative, with the ratio of performance to inter-quartile distance (RPIQ) of 1.85. Regarding the full-spectrum modeling for SOC, the CNN outperformed PLSR and Cubist models, irrespective of raw absorbance or eight FODs; the best-performing CNN model was achieved by 1.25-order derivative (validation RPIQ~=~6.33). It can be concluded that accurate estimation of SOC using large and diverse MIR spectral library at the global scale combined with deep-learning CNN model is feasible. This global-scale database is extremely valuable for us to deal with the shortage of soil data and to monitor the soils at different geographical scales.},
  keywords = {Deep learning,Fractional-order derivative,Mid-infrared spectroscopy,Soil monitoring,Soil spectral library},
  file = {C\:\\Users\\sjelic\\Zotero\\storage\\VM64H96Z\\Hong et al. - 2024 - Potential of globally distributed topsoil mid-infr.pdf;C\:\\Users\\sjelic\\Zotero\\storage\\JENIMDBF\\S0341816223007191.html}
}

@article{hsuComparisonIntegratedClustering2015,
  title = {Comparison of Integrated Clustering Methods for Accurate and Stable Prediction of Building Energy Consumption Data},
  author = {Hsu, David},
  date = {2015-12-15},
  journaltitle = {Applied Energy},
  shortjournal = {Applied Energy},
  volume = {160},
  pages = {153--163},
  issn = {0306-2619},
  doi = {10.1016/j.apenergy.2015.08.126},
  url = {https://www.sciencedirect.com/science/article/pii/S0306261915010624},
  abstract = {Clustering methods are often used to model energy consumption for two reasons. First, clustering is often used to process data and to improve the predictive accuracy of subsequent energy models. Second, stable clusters that are reproducible with respect to non-essential changes can be used to group, target, and interpret observed subjects. However, it is well known that clustering methods are highly sensitive to the choice of algorithms and variables. This can lead to misleading assessments of predictive accuracy and mis-interpretation of clusters in policymaking. This paper therefore introduces two methods to the modeling of energy consumption in buildings: clusterwise regression, also known as latent class regression, which integrates clustering and regression simultaneously; and cluster validation methods to measure stability. Using a large dataset of multifamily buildings in New York City, clusterwise regression is compared to common two-stage algorithms that use K-means and model-based clustering with linear regression. Predictive accuracy is evaluated using 20-fold cross validation, and the stability of the perturbed clusters is measured using the Jaccard coefficient. These results show that there seems to be an inherent tradeoff between prediction accuracy and cluster stability. This paper concludes by discussing which clustering methods may be appropriate for different analytical purposes.},
  keywords = {Buildings,Cluster stability,Cluster-wise regression,Energy consumption,Latent class regression,Prediction accuracy},
  file = {C\:\\Users\\sjelic\\Zotero\\storage\\YM8SHNPM\\Hsu - 2015 - Comparison of integrated clustering methods for ac.pdf;C\:\\Users\\sjelic\\Zotero\\storage\\A2VBTQDF\\S0306261915010624.html}
}

@article{hsuIdentifyingKeyVariables2015,
  title = {Identifying Key Variables and Interactions in Statistical Models of~Building Energy Consumption Using Regularization},
  author = {Hsu, David},
  date = {2015-04-01},
  journaltitle = {Energy},
  shortjournal = {Energy},
  volume = {83},
  pages = {144--155},
  issn = {0360-5442},
  doi = {10.1016/j.energy.2015.02.008},
  url = {https://www.sciencedirect.com/science/article/pii/S0360544215001590},
  abstract = {Statistical models can only be as good as the data put into them. Data about energy consumption continues to grow, particularly its non-technical aspects, but these variables are often interpreted differently among disciplines, datasets, and contexts. Selecting key variables and interactions is therefore an important step in achieving more accurate predictions, better interpretation, and identification of key subgroups for further analysis. This paper therefore makes two main contributions to the modeling and analysis of energy consumption of buildings. First, it introduces regularization, also known as penalized regression, for principled selection of variables and interactions. Second, this approach is demonstrated by application to a comprehensive dataset of energy consumption for commercial office and multifamily buildings in New York City. Using cross-validation, this paper finds that a newly-developed method, hierarchical group-lasso regularization, significantly outperforms ridge, lasso, elastic net and ordinary least squares approaches in terms of prediction accuracy; develops a parsimonious model for large New York City buildings; and identifies several interactions between technical and non-technical parameters for further analysis, policy development and targeting. This method is generalizable to other local contexts, and is likely to be useful for the modeling of other sectors of energy consumption as well.},
  keywords = {Buildings,Energy consumption,Statistical models,Variable selection},
  file = {C\:\\Users\\sjelic\\Zotero\\storage\\8JPUNFZJ\\Hsu - 2015 - Identifying key variables and interactions in stat.pdf;C\:\\Users\\sjelic\\Zotero\\storage\\W6U9AANU\\S0360544215001590.html}
}

@article{ibarakiEffectiveLocalSearch2005,
  title = {Effective {{Local Search Algorithms}} for {{Routing}} and {{Scheduling Problems}} with {{General Time-Window Constraints}}},
  author = {Ibaraki, T. and Imahori, S. and Kubo, M. and Masuda, T. and Uno, T. and Yagiura, M.},
  date = {2005},
  journaltitle = {Transportation Science},
  volume = {39},
  number = {2},
  eprint = {25769243},
  eprinttype = {jstor},
  pages = {206--232},
  publisher = {INFORMS},
  issn = {0041-1655},
  url = {https://www.jstor.org/stable/25769243},
  abstract = {We propose local search algorithms for the vehicle routing problem with soft time-window constraints. The time-window constraint for each customer is treated as a penalty function, which is very general in the sense that it can be nonconvex and discontinuous as long as it is piecewise linear. In our algorithm, we use local search to assign customers to vehicles and to find orders of customers for vehicles to visit. Our algorithm employs an advanced neighborhood, called the cyclic-exchange neighborhood, in addition to standard neighborhoods for the vehicle routing problem. After fixing the order of customers for a vehicle to visit, we must determine the optimal start times of processing at customers so that the total penalty is minimized. We show that this problem can be efficiently solved by using dynamic programming, which is then incorporated in our algorithm. We report computational results for various benchmark instances of the vehicle routing problem. The generality of time-window constraints allows us to handle a wide variety of scheduling problems. As an example, we mention in this paper an application to a production scheduling problem with inventory cost, and report computational results for real-world instances.},
  file = {C:\Users\sjelic\Zotero\storage\LJY54DCZ\Ibaraki et al. - 2005 - Effective Local Search Algorithms for Routing and Scheduling Problems with General Time-Window Const.pdf}
}

@inproceedings{ihlerBoundsQualityApproximate1991,
  title = {Bounds on the Quality of Approximate Solutions to the Group {{Steiner}} Problem},
  booktitle = {Graph-{{Theoretic Concepts}} in {{Computer Science}}},
  author = {Ihler, Edmund},
  editor = {Möhring, Rolf H.},
  date = {1991},
  series = {Lecture {{Notes}} in {{Computer Science}}},
  pages = {109--118},
  publisher = {Springer},
  location = {Berlin, Heidelberg},
  doi = {10.1007/3-540-53832-1_36},
  abstract = {The Group Steiner Problem (GSP) is a generalized version of the well known Steiner Problem. For an undirected, connected distance graph with groups of required vertices and Steiner vertices, GSP asks for a shortest connected subgraph, containing at least one vertex of each group. As the Steiner Problem is NP-hard, GSP is too, and we are interested in approximation algorithms. Efficient approximation algorithms have already been proposed, but nothing about the quality of any approximate solution is known so far. Especially for the VLSI design application of the problem, bounds on the quality of approximate solutions are of great importance.},
  isbn = {978-3-540-46310-8},
  langid = {english},
  keywords = {Approximation Algorithm,Input Graph,Polynomial Time,Span Tree,Steiner Tree}
}

@thesis{jelicFastApproximationAlgorithms2015,
  type = {info:eu-repo/semantics/doctoralThesis},
  title = {Fast approximation algorithms for connected set cover problem and related problems},
  author = {Jelić, Slobodan},
  date = {2015-05-28},
  institution = {University of Zagreb. Faculty of Science. Department of Mathematics},
  url = {https://urn.nsk.hr/urn:nbn:hr:217:865555},
  abstract = {A first part of the contribution in this thesis consists of approximation algorithms for Minimum Connected Set Cover problem (MCSC) which are published in [28]. First, we present a polylogarithmic approximation algorithm for MCSC problem that uses an approximation algorithm for the group Steiner tree problem (GST) in [37]. We also give a first approximation algorithm for the weighted version of the problem [28]. MCSC problem with demands is also considered. A demand of each element determines the smallest number of sets in the solution that covers considered element. A second part of the contribution is related to the approximation algorithm for GST problem where the size of each group is bounded by some constant. This special cases remains NPhard since it generalizes Steiner tree problem. In the thesis a constant approximation algorithm for this problem will be studied. We also give approximation algorithms for some related problems. A third part of the contribution consists of an adaptation of the algorithm for solving packing and covering linear programs, which is presented in [52], to the parallel computing platform that is supported by modern graphic NVidia chips with CUDA. Instead of updating a single entry in the primal and dual solution vector per iteration, we update a few randomly chosen entries. A part of the contribution is related to deterministic updates of many entries in the primal and dual solution vector which is more amenable for parallelization. Although it increases time per iteration, an update of multiple entries in the primal and dual vectors in one iteration will make the primal and dual vectors converge to the optimal solution much faster which leads to a fewer number of iteration. Approximation algorithms for GST problem use algorithms for solving relaxation of natural integer programming formulation [37] for GST. After integrality constraints are relaxed, a linear program becomes the covering linear program where the number of constraints is an exponential function of the input size. In the thesis, the fully polynomial time approximation scheme from [52, 62] will be adopted such that it approximates a solution of the LP relaxation of GST [51].},
  langid = {croatian}
}

@article{jelicGovernmentFormationProblem2018,
  title = {Government {{Formation Problem}}},
  author = {Jelić, Slobodan and Ševerdija, Domagoj},
  date = {2018-09-01},
  journaltitle = {Central European Journal of Operations Research},
  shortjournal = {Cent Eur J Oper Res},
  volume = {26},
  number = {3},
  pages = {659--672},
  issn = {1613-9178},
  doi = {10.1007/s10100-017-0505-8},
  url = {https://doi.org/10.1007/s10100-017-0505-8},
  abstract = {In addition to the same political and ideological attitudes, members of political parties can be interconnected at private and/or professional levels. They are considered as a part of one large social network. After democratic elections, the total effectiveness and stability of a government may depend on expertness and cooperability of its members. Our main goal is not to give a mechanism for pre-elective government formation, but to propose a model that decides what can be a good subset of candidates for positions in the new government. The decision is based on expertness of candidates and their interconnections in the social network. Inspired by the Team Formation Problem in a social network, we present a Government Formation Problem. We prove that this problem is NP-hard and give an algorithm based on integer linear programming formulation. In the experimental part, we test our algorithm on simulated data using the Gurobi MILP solver.},
  langid = {english},
  keywords = {Government Formation Problem,Integer linear programming,Social networks,Team Formation Problem}
}

@article{kalatzantonakisReinforcementLearningVariableNeighborhood2023a,
  title = {A Reinforcement Learning-{{Variable}} Neighborhood Search Method for the Capacitated {{Vehicle Routing Problem}}},
  author = {Kalatzantonakis, Panagiotis and Sifaleras, Angelo and Samaras, Nikolaos},
  date = {2023-03-01},
  journaltitle = {Expert Systems with Applications},
  shortjournal = {Expert Systems with Applications},
  volume = {213},
  pages = {118812},
  issn = {0957-4174},
  doi = {10.1016/j.eswa.2022.118812},
  url = {https://www.sciencedirect.com/science/article/pii/S0957417422018309},
  abstract = {Finding the best sequence of local search operators that yields the optimal performance of Variable Neighborhood Search (VNS) is an important open research question in the field of metaheuristics. This paper proposes a Reinforcement Learning method to address this question. We introduce a new hyperheuristic scheme, termed Bandit VNS, inspired by the Multi-Armed Bandit (MAB), a particular type of a single state reinforcement learning problem. In Bandit VNS, we utilize the General Variable Neighborhood Search metaheuristic and enhance it by a hyperheuristic strategy. We examine several variations of the Upper Confidence Bound algorithm to create a reliable strategy for adaptive neighborhood selection. Furthermore, we utilize Adaptive Windowing, a state of the art algorithm to estimate and detect changes in the data stream. Bandit VNS is designed for effective parallelization and encourages cooperation between agents to produce the best solution quality. We demonstrate this concept’s advantages in accuracy and speed by extensive experimentation using the Capacitated Vehicle Routing Problem. We compare the novel scheme’s performance against the conventional General Variable Neighborhood Search metaheuristic in terms of the CPU time and solution quality. The Bandit VNS method shows excellent results and reaches significantly higher performance metrics when applied to well-known benchmark instances. Our experiments show that, our approach achieves an improvement of more than 25\% in solution quality when compared to the General Variable Neighborhood Search method using standard library instances of medium and large size.},
  keywords = {Bandit Learning,Intelligent Optimization,Metaheuristics,Multi-Armed Bandits,Reinforcement Learning,Variable Neighborhood Search,Vehicle Routing Problem},
  file = {C\:\\Users\\sjelic\\Zotero\\storage\\WLLPDA4U\\Kalatzantonakis et al. - 2023 - A reinforcement learning-Variable neighborhood sea.pdf;C\:\\Users\\sjelic\\Zotero\\storage\\P9PJ8FB6\\S0957417422018309.html}
}

@incollection{kangClusterwiseRegressionUsing2009,
  title = {Clusterwise {{Regression Using Dirichlet Mixtures}}},
  booktitle = {Advances in {{Multivariate Statistical Methods}}},
  author = {Kang, Changku and Ghosal, Subhashis},
  date = {2009-06},
  series = {Statistical {{Science}} and {{Interdisciplinary Research}}},
  volume = {Volume 4},
  number = {Volume 4},
  pages = {305--325},
  publisher = {WORLD SCIENTIFIC},
  doi = {10.1142/9789812838247_0018},
  url = {https://www.worldscientific.com/doi/10.1142/9789812838247_0018},
  isbn = {978-981-283-823-0},
  file = {C:\Users\sjelic\Zotero\storage\NMGLZ7PR\Kang и Ghosal - 2009 - Clusterwise Regression Using Dirichlet Mixtures.pdf}
}

@article{karmitsaMissingValueImputation2022,
  title = {Missing {{Value Imputation}} via {{Clusterwise Linear Regression}}},
  author = {Karmitsa, Napsu and Taheri, Sona and Bagirov, Adil and Mäkinen, Pauliina},
  date = {2022-04},
  journaltitle = {IEEE Transactions on Knowledge and Data Engineering},
  volume = {34},
  number = {4},
  pages = {1889--1901},
  issn = {1558-2191},
  doi = {10.1109/TKDE.2020.3001694},
  url = {https://ieeexplore.ieee.org/document/9115681},
  abstract = {In this paper a new method of preprocessing incomplete data is introduced. The method is based on clusterwise linear regression and it combines two well-known approaches for missing value imputation: linear regression and clustering. The idea is to approximate missing values using only those data points that are somewhat similar to the incomplete data point. A similar idea is used also in clustering based imputation methods. Nevertheless, here the linear regression approach is used within each cluster to accurately predict the missing values, and this is done simultaneously to clustering. The proposed method is tested using some synthetic and real-world data sets and compared with other algorithms for missing value imputations. Numerical results demonstrate that this method produces the most accurate imputations in MCAR and MAR data sets with a clear structure and the percentages of missing data no more than 25 percent.},
  eventtitle = {{{IEEE Transactions}} on {{Knowledge}} and {{Data Engineering}}},
  keywords = {clusterwise linear regression,Data analysis,imputation,incomplete data,nonsmooth optimization,TV},
  file = {C\:\\Users\\sjelic\\Zotero\\storage\\HSJBRZFA\\Karmitsa et al. - 2022 - Missing Value Imputation via Clusterwise Linear Re.pdf;C\:\\Users\\sjelic\\Zotero\\storage\\Z5DX4U3D\\9115681.html}
}

@article{kimeldorfResultsTchebycheffianSpline1971,
  title = {Some Results on {{Tchebycheffian}} Spline Functions},
  author = {Kimeldorf, George and Wahba, Grace},
  date = {1971-01-01},
  journaltitle = {Journal of Mathematical Analysis and Applications},
  shortjournal = {Journal of Mathematical Analysis and Applications},
  volume = {33},
  number = {1},
  pages = {82--95},
  issn = {0022-247X},
  doi = {10.1016/0022-247X(71)90184-3},
  url = {https://www.sciencedirect.com/science/article/pii/0022247X71901843},
  abstract = {This report derives explicit solutions to problems involving Tchebycheffian spline functions. We use a reproducing kernel Hilbert space which depends on the smoothness criterion, but not on the form of the data, to solve explicitly Hermite-Birkhoff interpolation and smoothing problems. Sard's best approximation to linear functionals and smoothing with respect to linear inequality constraints are also discussed. Some of the results are used to show that spline interpolation and smoothing is equivalent to prediction and filtering on realizations of certain stochastic processes.},
  file = {C\:\\Users\\sjelic\\Zotero\\storage\\QV4ZIQTA\\Kimeldorf and Wahba - 1971 - Some results on Tchebycheffian spline functions.pdf;C\:\\Users\\sjelic\\Zotero\\storage\\CA448QLQ\\0022247X71901843.html}
}

@article{kiselevFlashlampBasedDevice2013,
  title = {A Flash-Lamp Based Device for Fluorescence Detection and Identification of Individual Pollen Grains},
  author = {Kiselev, Denis and Bonacina, Luigi and Wolf, Jean-Pierre},
  date = {2013-03},
  journaltitle = {The Review of Scientific Instruments},
  shortjournal = {Rev Sci Instrum},
  volume = {84},
  number = {3},
  eprint = {23556810},
  eprinttype = {pubmed},
  pages = {033302},
  issn = {1089-7623},
  doi = {10.1063/1.4793792},
  abstract = {We present a novel optical aerosol particle detector based on Xe flash lamp excitation and spectrally resolved fluorescence acquisition. We demonstrate its performances on three natural pollens acquiring in real-time scattering intensity at two wavelengths, sub-microsecond time-resolved scattering traces of the particles' passage in the focus, and UV-excited fluorescence spectra. We show that the device gives access to a rather specific detection of the bioaerosol particles.},
  langid = {english},
  keywords = {Aerosols,Cost-Benefit Analysis,Electronics,Environmental Monitoring,Equipment Design,Fluorescence,Lasers,Light,Normal Distribution,Pollen,Principal Component Analysis,Scattering Radiation,Spectrometry Fluorescence,Surface Properties,Ultraviolet Rays},
  file = {C:\Users\sjelic\Zotero\storage\UGHHCHKA\Kiselev et al. - 2013 - A flash-lamp based device for fluorescence detecti.pdf}
}

@article{kleijnenKrigingMetamodelingSimulation2009,
  title = {Kriging Metamodeling in Simulation: {{A}} Review},
  shorttitle = {Kriging Metamodeling in Simulation},
  author = {Kleijnen, Jack P. C.},
  date = {2009-02-01},
  journaltitle = {European Journal of Operational Research},
  shortjournal = {European Journal of Operational Research},
  volume = {192},
  number = {3},
  pages = {707--716},
  issn = {0377-2217},
  doi = {10.1016/j.ejor.2007.10.013},
  url = {https://www.sciencedirect.com/science/article/pii/S0377221707010090},
  abstract = {This article reviews Kriging (also called spatial correlation modeling). It presents the basic Kriging assumptions and formulas—contrasting Kriging and classic linear regression metamodels. Furthermore, it extends Kriging to random simulation, and discusses bootstrapping to estimate the variance of the Kriging predictor. Besides classic one-shot statistical designs such as Latin Hypercube Sampling, it reviews sequentialized and customized designs for sensitivity analysis and optimization. It ends with topics for future research.},
  keywords = {Design,Interpolation,Kriging,Metamodel,Optimization,Response surface},
  file = {C\:\\Users\\sjelic\\Zotero\\storage\\VP5FMG7P\\Kleijnen - 2009 - Kriging metamodeling in simulation A review.pdf;C\:\\Users\\sjelic\\Zotero\\storage\\2R4865DU\\S0377221707010090.html}
}

@article{kleinNearlyBestPossibleApproximation1995,
  title = {A {{Nearly Best-Possible Approximation Algorithm}} for {{Node-Weighted Steiner Trees}}},
  author = {Klein, P. and Ravi, R.},
  date = {1995-07-01},
  journaltitle = {Journal of Algorithms},
  shortjournal = {Journal of Algorithms},
  volume = {19},
  number = {1},
  pages = {104--115},
  issn = {0196-6774},
  doi = {10.1006/jagm.1995.1029},
  url = {https://www.sciencedirect.com/science/article/pii/S0196677485710292},
  abstract = {We give the first approximation algorithm for the node-weighted Steiner tree problem. Its performance guarantee is within a constant factor of the best possible unless P̃ ⊇ NP. (P̃ stands for the complexity class deterministic quasi-polynomial time, or DTIME[npolylog n].) Our algorithm generalizes to handle other network-design problems.},
  langid = {english},
  file = {C:\Users\sjelic\Zotero\storage\7F2ZIDVH\S0196677485710292.html}
}

@article{knightAsymptoticsLassoTypeEstimators2000,
  title = {Asymptotics for {{Lasso-Type Estimators}}},
  author = {Knight, Keith and Fu, Wenjiang},
  date = {2000},
  journaltitle = {The Annals of Statistics},
  volume = {28},
  number = {5},
  eprint = {2674097},
  eprinttype = {jstor},
  pages = {1356--1378},
  url = {http://www.jstor.org/stable/2674097},
  langid = {english},
  file = {C:\Users\sjelic\Zotero\storage\XK85RRU7\Knight and Fu - 2000 - Asymptotics for Lasso-Type Estimators.pdf}
}

@article{kochSolvingSteinerTree1998,
  title = {Solving {{Steiner}} Tree Problems in Graphs to Optimality},
  author = {Koch, T. and Martin, A.},
  date = {1998},
  journaltitle = {Networks},
  volume = {32},
  number = {3},
  pages = {207--232},
  issn = {1097-0037},
  doi = {10.1002/(SICI)1097-0037(199810)32:3<207::AID-NET5>3.0.CO;2-O},
  url = {https://onlinelibrary.wiley.com/doi/abs/10.1002/%28SICI%291097-0037%28199810%2932%3A3%3C207%3A%3AAID-NET5%3E3.0.CO%3B2-O},
  abstract = {In this paper, we present the implementation of a branch-and-cut algorithm for solving Steiner tree problems in graphs. Our algorithm is based on an integer programming formulation for directed graphs and comprises preprocessing, separation algorithms, and primal heuristics. We are able to solve nearly all problem instances discussed in the literature to optimality, including one problem that—to our knowledge—has not yet been solved. We also report on our computational experiences with some very large Steiner tree problems arising from the design of electronic circuits. All test problems are gathered in a newly introduced library called SteinLib that is accessible via the World Wide Web. © 1998 John Wiley \& Sons, Inc. Networks 32: 207–232, 1998},
  langid = {english},
  keywords = {branch-and-cut,cutting planes,reduction methods,Steiner tree,Steiner tree library},
  file = {C\:\\Users\\sjelic\\Zotero\\storage\\9RXG6AD6\\Koch and Martin - 1998 - Solving Steiner tree problems in graphs to optimal.pdf;C\:\\Users\\sjelic\\Zotero\\storage\\TZMBZFHB\\(SICI)1097-0037(199810)323207AID-NET53.0.html}
}

@online{konemannLMPLogApproximation2013,
  title = {An {{LMP O}}(Log n)-{{Approximation Algorithm}} for {{Node Weighted Prize Collecting Steiner Tree}}},
  author = {Könemann, Jochen and Sadeghian, Sina and Sanità, Laura},
  date = {2013-04-10},
  eprint = {1302.2127},
  eprinttype = {arXiv},
  eprintclass = {cs},
  doi = {10.48550/arXiv.1302.2127},
  url = {http://arxiv.org/abs/1302.2127},
  abstract = {In the node-weighted prize-collecting Steiner tree problem (NW-PCST) we are given an undirected graph \$G=(V,E)\$, non-negative costs \$c(v)\$ and penalties \$\textbackslash pi(v)\$ for each \$v \textbackslash in V\$. The goal is to find a tree \$T\$ that minimizes the total cost of the vertices spanned by \$T\$ plus the total penalty of vertices not in \$T\$. This problem is well-known to be set-cover hard to approximate. Moss and Rabani (STOC'01) presented a primal-dual Lagrangean-multiplier-preserving \$O(\textbackslash ln |V|)\$-approximation algorithm for this problem. We show a serious problem with the algorithm, and present a new, fundamentally different primal-dual method achieving the same performance guarantee. Our algorithm introduces several novel features to the primal-dual method that may be of independent interest.},
  pubstate = {prepublished},
  keywords = {Computer Science - Data Structures and Algorithms},
  file = {C\:\\Users\\sjelic\\Zotero\\storage\\VYCP4Y8N\\Könemann et al. - 2013 - An LMP O(log n)-Approximation Algorithm for Node W.pdf;C\:\\Users\\sjelic\\Zotero\\storage\\XYVFM9HC\\1302.html}
}

@online{kouwIntroductionDomainAdaptation2019,
  title = {An Introduction to Domain Adaptation and Transfer Learning},
  author = {Kouw, Wouter M. and Loog, Marco},
  date = {2019-01-14},
  eprint = {1812.11806},
  eprinttype = {arXiv},
  eprintclass = {cs, stat},
  url = {http://arxiv.org/abs/1812.11806},
  abstract = {In machine learning, if the training data is an unbiased sample of an underlying distribution, then the learned classification function will make accurate predictions for new samples. However, if the training data is not an unbiased sample, then there will be differences between how the training data is distributed and how the test data is distributed. Standard classifiers cannot cope with changes in data distributions between training and test phases, and will not perform well. Domain adaptation and transfer learning are sub-fields within machine learning that are concerned with accounting for these types of changes. Here, we present an introduction to these fields, guided by the question: when and how can a classifier generalize from a source to a target domain? We will start with a brief introduction into risk minimization, and how transfer learning and domain adaptation expand upon this framework. Following that, we discuss three special cases of data set shift, namely prior, covariate and concept shift. For more complex domain shifts, there are a wide variety of approaches. These are categorized into: importance-weighting, subspace mapping, domain-invariant spaces, feature augmentation, minimax estimators and robust algorithms. A number of points will arise, which we will discuss in the last section. We conclude with the remark that many open questions will have to be addressed before transfer learners and domain-adaptive classifiers become practical.},
  pubstate = {prepublished},
  keywords = {Computer Science - Computer Vision and Pattern Recognition,Computer Science - Machine Learning,Statistics - Machine Learning},
  file = {C\:\\Users\\sjelic\\Zotero\\storage\\RHW52ET5\\Kouw and Loog - 2019 - An introduction to domain adaptation and transfer .pdf;C\:\\Users\\sjelic\\Zotero\\storage\\GZ74UAIX\\1812.html}
}

@article{kucheryavskiyProcrustesCrossvalidationMultivariate2023,
  title = {Procrustes Cross-Validation of Multivariate Regression Models},
  author = {Kucheryavskiy, Sergey and Rodionova, Oxana and Pomerantsev, Alexey},
  date = {2023-05-15},
  journaltitle = {Analytica Chimica Acta},
  shortjournal = {Analytica Chimica Acta},
  volume = {1255},
  pages = {341096},
  issn = {0003-2670},
  doi = {10.1016/j.aca.2023.341096},
  url = {https://www.sciencedirect.com/science/article/pii/S0003267023003173},
  abstract = {A generalization of Procrustes Cross-Validation — recently introduced novel approach for validation of chemometric models — is proposed. The generalized approach is faster than its predecessor by several orders of magnitude and can be used for validation of a wider range of models. Furthermore, it provides new tools for exploring the heterogeneity of the dataset, quality of cross-validation splits, presence of outliers, etc. The paper describes methodological aspects of the generalized approach, based on using Procrustean rules, the mathematical background, as well as presents practical results obtained using real chemical datasets.},
  file = {C\:\\Users\\sjelic\\Zotero\\storage\\8YJIPAVW\\Kucheryavskiy et al. - 2023 - Procrustes cross-validation of multivariate regres.pdf;C\:\\Users\\sjelic\\Zotero\\storage\\M7FEA56N\\S0003267023003173.html}
}

@inproceedings{lappasFindingTeamExperts2009,
  title = {Finding a Team of Experts in Social Networks},
  booktitle = {Proceedings of the 15th {{ACM SIGKDD}} International Conference on {{Knowledge}} Discovery and Data Mining},
  author = {Lappas, Theodoros and Liu, Kun and Terzi, Evimaria},
  date = {2009-06-28},
  series = {{{KDD}} '09},
  pages = {467--476},
  publisher = {Association for Computing Machinery},
  location = {New York, NY, USA},
  doi = {10.1145/1557019.1557074},
  url = {https://doi.org/10.1145/1557019.1557074},
  abstract = {Given a task T, a pool of individuals X with different skills, and a social network G that captures the compatibility among these individuals, we study the problem of finding X, a subset of X, to perform the task. We call this the TEAM FORMATION problem. We require that members of X' not only meet the skill requirements of the task, but can also work effectively together as a team. We measure effectiveness using the communication cost incurred by the subgraph in G that only involves X'. We study two variants of the problem for two different communication-cost functions, and show that both variants are NP-hard. We explore their connections with existing combinatorial problems and give novel algorithms for their solution. To the best of our knowledge, this is the first work to consider the TEAM FORMATION problem in the presence of a social network of individuals. Experiments on the DBLP dataset show that our framework works well in practice and gives useful and intuitive results.},
  isbn = {978-1-60558-495-9},
  keywords = {graph algorithms,social networks,team formation}
}

@online{larssonChoiceNormalizationInfluences2025,
  title = {The {{Choice}} of {{Normalization Influences Shrinkage}} in {{Regularized Regression}}},
  author = {Larsson, Johan and Wallin, Jonas},
  date = {2025-01-21},
  eprint = {2501.03821},
  eprinttype = {arXiv},
  eprintclass = {stat},
  doi = {10.48550/arXiv.2501.03821},
  url = {http://arxiv.org/abs/2501.03821},
  abstract = {Regularized models are often sensitive to the scales of the features in the data and it has therefore become standard practice to normalize (center and scale) the features before fitting the model. But there are many different ways to normalize the features and the choice may have dramatic effects on the resulting model. In spite of this, there has so far been no research on this topic. In this paper, we begin to bridge this knowledge gap by studying normalization in the context of lasso, ridge, and elastic net regression. We focus on normal and binary features and show that the class balances of binary features directly influences the regression coefficients and that this effect depends on the combination of normalization and regularization methods used. We demonstrate that this effect can be mitigated by scaling binary features with their variance in the case of the lasso and standard deviation in the case of ridge regression, but that this comes at the cost of increased variance. For the elastic net, we show that scaling the penalty weights, rather than the features, can achieve the same effect. Finally, we also tackle mixes of binary and normal features as well as interactions and provide some initial results on how to normalize features in these cases.},
  pubstate = {prepublished},
  keywords = {Computer Science - Machine Learning,Statistics - Machine Learning,Statistics - Methodology},
  file = {C\:\\Users\\sjelic\\Zotero\\storage\\W76S52GN\\Larsson and Wallin - 2025 - The Choice of Normalization Influences Shrinkage i.pdf;C\:\\Users\\sjelic\\Zotero\\storage\\35RQWXZ4\\2501.html}
}

@article{lauMathematicalProgrammingApproach1999,
  title = {A Mathematical Programming Approach to Clusterwise Regression Model and Its Extensions},
  author = {Lau, Kin-nam and Leung, Pui-lam and Tse, Ka-kit},
  date = {1999-08-01},
  journaltitle = {European Journal of Operational Research},
  shortjournal = {European Journal of Operational Research},
  volume = {116},
  number = {3},
  pages = {640--652},
  issn = {0377-2217},
  doi = {10.1016/S0377-2217(98)00052-6},
  url = {https://www.sciencedirect.com/science/article/pii/S0377221798000526},
  abstract = {The clusterwise regression model is used to perform cluster analysis within a regression framework. While the traditional regression model assumes the regression coefficient (β) to be identical for all subjects in the sample, the clusterwise regression model allows β to vary with subjects of different clusters. Since the cluster membership is unknown, the estimation of the clusterwise regression is a tough combinatorial optimization problem. In this research, we propose a “Generalized Clusterwise Regression Model” which is formulated as a mathematical programming (MP) problem. A nonlinear programming procedure (with linear constraints) is proposed to solve the combinatorial problem and to estimate the cluster membership and β simultaneously. Moreover, by integrating the cluster analysis with the discriminant analysis, a clusterwise discriminant model is developed to incorporate parameter heterogeneity into the traditional discriminant analysis. The cluster membership and discriminant parameters are estimated simultaneously by another nonlinear programming model.},
  keywords = {Cluster analysis,Clusterwise regression,Discriminant analysis,Mathematical programming,Multivariate statistics},
  file = {C\:\\Users\\sjelic\\Zotero\\storage\\BA9HHECP\\Lau et al. - 1999 - A mathematical programming approach to clusterwise.pdf;C\:\\Users\\sjelic\\Zotero\\storage\\MU45VAXZ\\S0377221798000526.html}
}

@article{leeExactPostselectionInference2016,
  title = {Exact Post-Selection Inference, with Application to the Lasso},
  author = {Lee, Jason D. and Sun, Dennis L. and Sun, Yuekai and Taylor, Jonathan E.},
  date = {2016-06-01},
  journaltitle = {The Annals of Statistics},
  shortjournal = {Ann. Statist.},
  volume = {44},
  number = {3},
  issn = {0090-5364},
  doi = {10.1214/15-AOS1371},
  url = {https://projecteuclid.org/journals/annals-of-statistics/volume-44/issue-3/Exact-post-selection-inference-with-application-to-the-lasso/10.1214/15-AOS1371.full},
  langid = {english},
  file = {C:\Users\sjelic\Zotero\storage\YW3Y26VJ\Lee et al. - 2016 - Exact post-selection inference, with application t.pdf}
}

@article{liClusterwiseFunctionalLinear2021,
  title = {Clusterwise Functional Linear Regression Models},
  author = {Li, Ting and Song, Xinyuan and Zhang, Yingying and Zhu, Hongtu and Zhu, Zhongyi},
  date = {2021-06-01},
  journaltitle = {Computational Statistics \& Data Analysis},
  shortjournal = {Computational Statistics \& Data Analysis},
  volume = {158},
  pages = {107192},
  issn = {0167-9473},
  doi = {10.1016/j.csda.2021.107192},
  url = {https://www.sciencedirect.com/science/article/pii/S0167947321000268},
  abstract = {Classical clusterwise linear regression is a useful method for investigating the relationship between scalar predictors and scalar responses with heterogeneous variation of regression patterns for different subgroups of subjects. This paper extends the classical clusterwise linear regression to incorporate multiple functional predictors by representing the functional coefficients in terms of a functional principal component basis. We estimate the functional principal component coefficients based on M-estimation and K-means clustering algorithm, which can classify the data into clusters and estimate clusterwise coefficients simultaneously. One advantage of the proposed method is that it is robust and flexible by adopting a general loss function, which can be broadly applied to mean regression, median regression, quantile regression and robust mean regression. A Bayesian information criterion is proposed to select the unknown number of groups and shown to be consistent in model selection. We also obtain the convergence rate of the set of estimators to the set of true coefficients for all clusters. Simulation studies and real data analysis show that the proposed method is easily implemented, and it consequently improves previous works and also requires much less computing burden than existing methods.},
  langid = {english},
  keywords = {Bayesian information criterion consistency,M-estimation,Subgroup analysis},
  file = {C\:\\Users\\sjelic\\Zotero\\storage\\4EIU6Q6T\\Li et al. - 2021 - Clusterwise functional linear regression models.pdf;C\:\\Users\\sjelic\\Zotero\\storage\\I663T4V6\\S0167947321000268.html}
}

@inproceedings{liEfficientProgressiveGroup2016,
  title = {Efficient and {{Progressive Group Steiner Tree Search}}},
  booktitle = {Proceedings of the 2016 {{International Conference}} on {{Management}} of {{Data}}},
  author = {Li, Rong-Hua and Qin, Lu and Yu, Jeffrey Xu and Mao, Rui},
  date = {2016-06-14},
  series = {{{SIGMOD}} '16},
  pages = {91--106},
  publisher = {Association for Computing Machinery},
  location = {New York, NY, USA},
  doi = {10.1145/2882903.2915217},
  url = {https://doi.org/10.1145/2882903.2915217},
  abstract = {The Group Steiner Tree (GST) problem is a fundamental problem in database area that has been successfully applied to keyword search in relational databases and team search in social networks. The state-of-the-art algorithm for the GST problem is a parameterized dynamic programming (DP) algorithm, which finds the optimal tree in O(3kn+2k(n log n + m)) time, where k is the number of given groups, m and n are the number of the edges and nodes of the graph respectively. The major limitations of the parameterized DP algorithm are twofold: (i) it is intractable even for very small values of k (e.g., k=8) in large graphs due to its exponential complexity, and (ii) it cannot generate a solution until the algorithm has completed its entire execution. To overcome these limitations, we propose an efficient and progressive GST algorithm in this paper, called PrunedDP. It is based on newly-developed optimal-tree decomposition and conditional tree merging techniques. The proposed algorithm not only drastically reduces the search space of the parameterized DP algorithm, but it also produces progressively-refined feasible solutions during algorithm execution. To further speed up the PrunedDP algorithm, we propose a progressive A*-search algorithm, based on several carefully-designed lower-bounding techniques. We conduct extensive experiments to evaluate our algorithms on several large scale real-world graphs. The results show that our best algorithm is not only able to generate progressively-refined feasible solutions, but it also finds the optimal solution with at least two orders of magnitude acceleration over the state-of-the-art algorithm, using much less memory.},
  isbn = {978-1-4503-3531-7},
  keywords = {a *-search algorithm,DP,group steiner tree}
}

@article{limBerthPlanningProblem1998,
  title = {The Berth Planning Problem},
  author = {Lim, Andrew},
  date = {1998-03-01},
  journaltitle = {Operations Research Letters},
  shortjournal = {Operations Research Letters},
  volume = {22},
  number = {2},
  pages = {105--110},
  issn = {0167-6377},
  doi = {10.1016/S0167-6377(98)00010-8},
  url = {https://www.sciencedirect.com/science/article/pii/S0167637798000108},
  abstract = {Singapore has one of the busiest ports in the world. Berth planning is one of the problems faced by planners. This paper studies the berth planning problem. We first formulate the problem and show that it is NP-Complete. We transform the berthing problem to a restricted form of the two-dimensional packing problem, use a graph theoretical representation to capture the problem succinctly, and propose an effective heuristic for the problem. Experimental results show that our heuristic performs well on historical test data.},
  keywords = {Berth planning,Graphs,Heuristic,NP-Complete},
  file = {C\:\\Users\\sjelic\\Zotero\\storage\\89HRWYK5\\1-s2.0-S0167637798000108-main.pdf;C\:\\Users\\sjelic\\Zotero\\storage\\9ZKS4WB5\\Lim - 1998 - The berth planning problem1.pdf;C\:\\Users\\sjelic\\Zotero\\storage\\VRZLKAWL\\S0167637798000108.html}
}

@incollection{liPolynomialIntegralityGap2022,
  title = {Polynomial {{Integrality Gap}} of {{Flow LP}} for {{Directed Steiner Tree}}},
  booktitle = {Proceedings of the 2022 {{Annual ACM-SIAM Symposium}} on {{Discrete Algorithms}} ({{SODA}})},
  author = {Li, Shi and Laekhanukit, Bundit},
  date = {2022-01},
  series = {Proceedings},
  pages = {3230--3236},
  publisher = {{Society for Industrial and Applied Mathematics}},
  doi = {10.1137/1.9781611977073.126},
  url = {https://epubs.siam.org/doi/abs/10.1137/1.9781611977073.126},
  abstract = {In the Directed Steiner Tree (DST) problem, we are given a directed graph G = (V, E) on n vertices with edge-costs , a root vertex r, and a set K of k terminals. The goal is to find a minimum-cost subgraph of G that contains a path from r to every terminal t ∊ k. DST has been a notorious problem for decades as there is a large gap between the best-known polynomial-time approximation ratio of O(k∊) for any constant ∊ {$>$} 0, and the best quasi-polynomial-time approximation ratio of . Towards understanding this gap, we study the integrality gap of the standard flow LP relaxation for the problem. We show that the LP has an integrality gap polynomial in n. Previously, the integrality gap LP is only known to be  [Halperin et al., SODA'03 \& SIAM J. Comput.] and  [Zosin-Khuller, SODA'02] in some instance with . Our result gives the first known lower bound on the integrality gap of this standard LP that is polynomial in n, the number of vertices. Consequently, we rule out the possibility of developing a poly-logarithmic approximation algorithm for the problem based on the flow LP relaxation.},
  file = {C:\Users\sjelic\Zotero\storage\K7XCWNCE\Li and Laekhanukit - 2022 - Polynomial Integrality Gap of Flow LP for Directed.pdf}
}

@article{liuUsingSentinel1Sentinel22022,
  title = {Using {{Sentinel-1}}, {{Sentinel-2}}, and {{Planet}} Satellite Data to Map Field-Level Tillage Practices in Smallholder Systems},
  author = {Liu, Yin and Rao, Preeti and Zhou, Weiqi and Singh, Balwinder and Srivastava, Amit K. and Poonia, Shishpal P. and Van Berkel, Derek and Jain, Meha},
  editor = {Nair, Jaishanker Raghunathan},
  date = {2022-11-28},
  journaltitle = {PLOS ONE},
  shortjournal = {PLoS ONE},
  volume = {17},
  number = {11},
  pages = {e0277425},
  issn = {1932-6203},
  doi = {10.1371/journal.pone.0277425},
  url = {https://dx.plos.org/10.1371/journal.pone.0277425},
  abstract = {Remote sensing can be used to map tillage practices at large spatial and temporal scales. However, detecting such management practices in smallholder systems is challenging given that the size of fields is smaller than historical readily-available satellite imagery. In this study we used newer, higher-resolution satellite data from Sentinel-1, Sentinel-2, and Planet to map tillage practices in the Eastern Indo-Gangetic Plains in India. We specifically tested the classification performance of single sensor and multiple sensor random forest models, and the impact of spatial, temporal, or spectral resolution on classification accuracy. We found that when considering a single sensor, the model that used Planet imagery (3 m) had the highest classification accuracy (86.55\%) while the model that used Sentinel-1 data (10 m) had the lowest classification accuracy (62.28\%). When considering sensor combinations, the model that used data from all three sensors achieved the highest classification accuracy (87.71\%), though this model was not statistically different from the Planet only model when considering 95\% confidence intervals from bootstrap analyses. We also found that high levels of accuracy could be achieved by only using imagery from the sowing period. Considering the impact of spatial, temporal, and spectral resolution on classification accuracy, we found that improved spatial resolution from Planet contributed the most to improved classification accuracy. Overall, it is possible to use readily-available, high spatial resolution satellite data to map tillage practices of smallholder farms, even in heterogeneous systems with small field sizes.},
  langid = {english},
  file = {C:\Users\sjelic\Zotero\storage\KTE6CHVQ\Liu et al. - 2022 - Using Sentinel-1, Sentinel-2, and Planet satellite.pdf}
}

@article{longMethodsApplicationsClusterwise2023,
  title = {Methods and {{Applications}} of {{Clusterwise Linear Regression}}: {{A Survey}} and {{Comparison}}},
  shorttitle = {Methods and {{Applications}} of {{Clusterwise Linear Regression}}},
  author = {Long, Qiang and Bagirov, Adil and Taheri, Sona and Sultanova, Nargiz and Wu, Xue},
  date = {2023-06-30},
  journaltitle = {ACM Transactions on Knowledge Discovery from Data},
  shortjournal = {ACM Trans. Knowl. Discov. Data},
  volume = {17},
  number = {3},
  pages = {1--54},
  issn = {1556-4681, 1556-472X},
  doi = {10.1145/3550074},
  url = {https://dl.acm.org/doi/10.1145/3550074},
  abstract = {Clusterwise linear regression (CLR) is a well-known technique for approximating a data using more than one linear function. It is based on the combination of clustering and multiple linear regression methods. This article provides a comprehensive survey and comparative assessments of CLR including model formulations, description of algorithms, and their performance on small to large-scale synthetic and real-world datasets. Some applications of the CLR algorithms and possible future research directions are also discussed.},
  langid = {english},
  file = {C:\Users\sjelic\Zotero\storage\I6QFU4CQ\Long и сар. - 2023 - Methods and Applications of Clusterwise Linear Reg.pdf}
}

@article{maAdaptiveAdversarialDomain2021,
  title = {An Adaptive Adversarial Domain Adaptation Approach for Corn Yield Prediction},
  author = {Ma, Yuchi and Zhang, Zhou and Yang, Hsiuhan Lexie and Yang, Zhengwei},
  date = {2021-08-01},
  journaltitle = {Computers and Electronics in Agriculture},
  shortjournal = {Computers and Electronics in Agriculture},
  volume = {187},
  pages = {106314},
  issn = {0168-1699},
  doi = {10.1016/j.compag.2021.106314},
  url = {https://www.sciencedirect.com/science/article/pii/S0168169921003318},
  abstract = {Recently, statistical machine learning and deep learning methods have been widely explored for corn yield prediction. Though successful, machine learning models generated within a specific spatial domain often lose their validity when directly applied to new regions. To address this issue, we designed an unsupervised adaptive domain adversarial neural network (ADANN). Specifically, through domain adversarial training, the ADANN model reduced the impact of domain shift by projecting data from different domains into the same subspace. Also, the ADANN model was designed to be trained in an adaptive way, which guaranteed the model can learn the domain-invariant features and perform accurate yield prediction simultaneously. Informative variables including time-series vegetation indices and sequential weather observations were first collected from multiple data sources and aggregated to the county level. Then, we trained the ADANN model with the extracted features and corresponding reported county-level corn yield from the U.S. Department of Agriculture (USDA). Finally, the trained model was evaluated in four testing years 2016–2019. The U.S. corn belt was used as the study area and counties under study were grouped into two diverse ecological regions. The experimental results showed that the developed ADANN model had better performance than three other state-of-the-art machine learning models in both local experiments (train and test in the same region) and transfer experiments (train and test in different regions). As the first study using adversarial learning for crop yield prediction, this research demonstrates a novel solution for improving model transferability on crop yield prediction.},
  langid = {english},
  keywords = {Adversarial training,Domain adaptation,Remote sensing,Transfer learning,Yield prediction},
  file = {C\:\\Users\\sjelic\\Zotero\\storage\\J6U7TI6C\\!!document.pdf;C\:\\Users\\sjelic\\Zotero\\storage\\8CMGDN3I\\S0168169921003318.html}
}

@article{malekahmadiIntegratedContinuousBerth2020,
  title = {Integrated Continuous Berth Allocation and Quay Crane Assignment and Scheduling Problem with Time-Dependent Physical Constraints in Container Terminals},
  author = {Malekahmadi, Amirsalar and Alinaghian, Mehdi and Hejazi, Seyed Reza and Assl Saidipour, Mohammad Ali},
  date = {2020-09-01},
  journaltitle = {Computers \& Industrial Engineering},
  shortjournal = {Computers \& Industrial Engineering},
  volume = {147},
  pages = {106672},
  issn = {0360-8352},
  doi = {10.1016/j.cie.2020.106672},
  url = {https://www.sciencedirect.com/science/article/pii/S036083522030406X},
  abstract = {This paper presents an integer programming model for integrated continuous berth allocation, quay crane assignment and quay crane scheduling problem (BACASP) in container terminals where berthing possibility depends on water depth and tide conditions. The proposed model also considers the safe distance between quay cranes and the fact that they cannot cross each other on the rails. Given the NP-Hard complexity of the proposed model, a particle swarm optimization (PSO) based meta-heuristic called the random topology particle swarm optimization algorithm (RTPSO) is developed for solving its large-size instances. To evaluate the performance of the developed RTPSO, its results are compared with the results of the exact solution and the basic PSO. The results illustrate the better performance of the proposed random topology particle swarm optimization algorithm in terms of accuracy and computational time.},
  keywords = {Berth allocation,Integrated model,Quay crane assignment,Quay crane scheduling,Tide,Time dependence},
  file = {C\:\\Users\\sjelic\\Zotero\\storage\\FS9T99KI\\Malekahmadi et al. - 2020 - Integrated continuous berth allocation and quay cr.pdf;C\:\\Users\\sjelic\\Zotero\\storage\\SFPMW2YS\\S036083522030406X.html}
}

@article{manwaniKplaneRegression2015,
  title = {K-Plane Regression},
  author = {Manwani, Naresh and Sastry, P. S.},
  date = {2015-01-20},
  journaltitle = {Information Sciences},
  shortjournal = {Information Sciences},
  volume = {292},
  pages = {39--56},
  issn = {0020-0255},
  doi = {10.1016/j.ins.2014.08.058},
  url = {https://www.sciencedirect.com/science/article/pii/S0020025514008639},
  abstract = {In this paper, we present a novel algorithm for piecewise linear regression which can learn continuous as well as discontinuous piecewise linear functions. The main idea is to repeatedly partition the data and learn a linear model in each partition. The proposed algorithm is similar in spirit to k-means clustering algorithm. We show that our algorithm can also be viewed as a special case of an EM algorithm for maximum likelihood estimation under a reasonable probability model. We empirically demonstrate the effectiveness of our approach by comparing its performance with that of the state of art algorithms on various datasets.},
  langid = {english},
  keywords = {Cluster-wise linear regression,Expectation maximization,Mixture of experts,Piecewise linear regression},
  file = {C:\Users\sjelic\Zotero\storage\WUIL546D\Manwani и Sastry - 2015 - K-plane regression.pdf}
}

@article{matavuljDomainAdaptationImproving2025,
  title = {Domain Adaptation for Improving Automatic Airborne Pollen Classification with Expert-Verified Measurements},
  author = {Matavulj, Predrag and Jelic, Slobodan and Severdija, Domagoj and Brdar, Sanja and Radovanovic, Milos and Tesendic, Danijela and Sikoparija, Branko},
  date = {2025-02-10},
  journaltitle = {Applied Intelligence},
  shortjournal = {Appl Intell},
  volume = {55},
  number = {6},
  pages = {430},
  issn = {1573-7497},
  doi = {10.1007/s10489-024-06021-9},
  url = {https://doi.org/10.1007/s10489-024-06021-9},
  abstract = {This study presents a novel approach to enhance the accuracy of automatic classification systems for airborne pollen particles by integrating domain adaptation techniques. Our method incorporates expert-verified measurements into the convolutional neural network (CNN) training process to address the discrepancy between laboratory test data and real-world environmental measurements. We systematically fine-tuned CNN models, initially developed on standard reference datasets, with these expert-verified measurements. A comprehensive exploration of hyperparameters was conducted to optimize the CNN models, ensuring their robustness and adaptability across various environmental conditions and pollen types. Empirical results indicate a significant improvement, evidenced by a 22.52\% increase in correlation and a 38.05\% reduction in standard deviation across 29 cases of different pollen classes over multiple study years. This research highlights the potential of domain adaptation techniques in environmental monitoring, particularly in contexts where the integrity and representativeness of reference datasets are difficult to verify.},
  langid = {english},
  keywords = {Artificial Intelligence,Classification,Domain adaptation,Neural networks,Pollen,Real-time},
  file = {C:\Users\sjelic\Zotero\storage\XN6QRNPK\Matavulj et al. - 2025 - Domain adaptation for improving automatic airborne pollen classification with expert-verified measur.pdf}
}

@article{matijevicGeneralVariableNeighborhood2022,
  title = {General Variable Neighborhood Search Approach to Group Steiner Tree Problem},
  author = {Matijević, Luka and Jelić, Slobodan and Davidović, Tatjana},
  date = {2022-07-07},
  journaltitle = {Optimization Letters},
  shortjournal = {Optim Lett},
  issn = {1862-4480},
  doi = {10.1007/s11590-022-01904-7},
  url = {https://doi.org/10.1007/s11590-022-01904-7},
  abstract = {In this paper, we consider the Group Steiner Tree (GST) problem that can be stated as follows: For a given non-negative edge weighted graph \$\$G = (V, E)\$\$, an integer k, and the corresponding family \$\$g\_1, \textbackslash ldots , g\_k\$\$containing non-empty subsets of V called groups, we need to find a minimum cost tree \$\$T = (V\_T, E\_T)\$\$where \$\$V\_T \textbackslash subseteq V\$\$and \$\$E\_T\textbackslash subseteq E\$\$that spans at least one vertex from each of the groups. Numerous applications of this NP-hard problem initiated researchers to study it from both theoretical and algorithmic aspects. One of the challenges is to provide a good heuristic solution within the reasonable amount of CPU time. We propose the application of metaheuristic framework based on Variable Neighborhood Search (VNS) and related approaches. One of our main objectives is to find a neighborhood structure that ensures efficient implementation. We develop Variable Neighborhood Descend (VND) algorithm that can be the main ingredient of several local search approaches. Experimental evaluation involves comparison of our heuristic to exact approach based on Integer Linear Programming solvers and other metaheuristic approaches, such as genetic algorithm. The obtained results show that the proposed method always outperforms genetic algorithm. Exact method is outperformed in the case of instances with large number of groups.},
  langid = {english},
  keywords = {Heuristic methods,Local search,Optimization on graphs,Stochastic search,Suboptimal solutions},
  file = {C:\Users\sjelic\Zotero\storage\X4NY2ET9\Matijević et al. - 2022 - General variable neighborhood search approach to g.pdf}
}

@article{megiddoComplexityLocatingLinear1982a,
  title = {On the Complexity of Locating Linear Facilities in the Plane},
  author = {Megiddo, Nimrod and Tamir, Arie},
  date = {1982-11-01},
  journaltitle = {Operations Research Letters},
  shortjournal = {Operations Research Letters},
  volume = {1},
  number = {5},
  pages = {194--197},
  issn = {0167-6377},
  doi = {10.1016/0167-6377(82)90039-6},
  url = {https://www.sciencedirect.com/science/article/pii/0167637782900396},
  abstract = {We consider the computational complexity of linear facility location problems in the plane, i.e., given n demand points, one wishes to find r lines so as to minimize a certain objective-function reflecting the need of the points to be close to the lines. It is shown that it is NP-hard to find r lines so as to minimize any isotone function of the distances between given points and their respective nearest lines. The proofs establish NP-hardness in the strong sense. The results also apply to the situation where the demand is represented by r lines and the facilities by n single points.},
  langid = {english},
  keywords = {NP-complete,p-line center,p-line median,planar location,strongly NP-complete},
  file = {C\:\\Users\\sjelic\\Zotero\\storage\\KCEPHA23\\complexity of locating linear facilities.pdf;C\:\\Users\\sjelic\\Zotero\\storage\\HFV5H8SK\\0167637782900396.html}
}

@article{micci-barrecaPreprocessingSchemeHighcardinality2001,
  title = {A Preprocessing Scheme for High-Cardinality Categorical Attributes in Classification and Prediction Problems},
  author = {Micci-Barreca, Daniele},
  date = {2001-07-01},
  journaltitle = {SIGKDD Explor. Newsl.},
  volume = {3},
  number = {1},
  pages = {27--32},
  issn = {1931-0145},
  doi = {10.1145/507533.507538},
  url = {https://doi.org/10.1145/507533.507538},
  abstract = {Categorical data fields characterized by a large number of distinct values represent a serious challenge for many classification and regression algorithms that require numerical inputs. On the other hand, these types of data fields are quite common in real-world data mining applications and often contain potentially relevant information that is difficult to represent for modeling purposes.This paper presents a simple preprocessing scheme for high-cardinality categorical data that allows this class of attributes to be used in predictive models such as neural networks, linear and logistic regression. The proposed method is based on a well-established statistical method (empirical Bayes) that is straightforward to implement as an in-database procedure. Furthermore, for categorical attributes with an inherent hierarchical structure, like ZIP codes, the preprocessing scheme can directly leverage the hierarchy by blending statistics at the various levels of aggregation.While the statistical methods discussed in this paper were first introduced in the mid 1950's, the use of these methods as a preprocessing step for complex models, like neural networks, has not been previously discussed in any literature.}
}

@article{mirzaeitalarposhtiDigitalSoilTexture2022,
  title = {Digital {{Soil Texture Mapping}} and {{Spatial Transferability}} of {{Machine Learning Models Using Sentinel-1}}, {{Sentinel-2}}, and {{Terrain-Derived Covariates}}},
  author = {Mirzaeitalarposhti, Reza and Shafizadeh-Moghadam, Hossein and Taghizadeh-Mehrjardi, Ruhollah and Demyan, Michael Scott},
  date = {2022-11-22},
  journaltitle = {Remote Sensing},
  shortjournal = {Remote Sensing},
  volume = {14},
  number = {23},
  pages = {5909},
  issn = {2072-4292},
  doi = {10.3390/rs14235909},
  url = {https://www.mdpi.com/2072-4292/14/23/5909},
  abstract = {Soil texture is an important property that controls the mobility of the water and nutrients in soil. This study examined the capability of machine learning (ML) models in estimating soil texture fractions using different combinations of remotely sensed data from Sentinel-1 (S1), Sentinel-2 (S2), and terrain-derived covariates (TDC) across two contrasting agroecological regions in Southwest Germany, Kraichgau and the Swabian Alb. Importantly, we tested the predictive power of three different ML models: the random forest (RF), the support vector machine (SVM), and extreme gradient boosting (XGB) coupled with the remote sensing data covariates. As expected, ML model performance was not consistent regarding the input covariates, soil texture fractions, and study regions. For example, in the Swabian Alb, the SVM model performed the best for the sand content with S2 + TDC (RMSE = 3.63\%, R2 = 0.42), and XGB best predicted the clay content with S1 + S2 + TDC (RMSE = 6.84\%, R2 = 0.64). In Kraichgau, the best models for sand (RMSE = 7.54\%, R2 = 0.79) and clay contents (RMSE = 6.14\%, R2 = 0.48) were obtained using XGB and SVM, respectively. Moreover, the results indicated that TDC were critical in estimating soil texture fractions, especially in Kraichgau, which indicated that topography plays an important role in defining the spatial distribution of soil properties. In contrast, the contribution of remote sensing data better predicted the silt and clay content in the Swabian Alb. The transferability of a region-specific model to the other region was low as indicated by poor predictive performance. The resulting soil-texture-fraction maps could be a significant source of information for efficient land resource management and environmental monitoring. Nonetheless, further research to evaluate the added value of the Sentinel imagery and to better analyze the spatial transferability of machine learning models is highly recommended.},
  langid = {english},
  file = {C:\Users\sjelic\Zotero\storage\ZJA893RL\Mirzaeitalarposhti et al. - 2022 - Digital Soil Texture Mapping and Spatial Transfera.pdf}
}

@article{mohammadiSourcefreeUnsupervisedDomain2024,
  title = {A Source-Free Unsupervised Domain Adaptation Method for Cross-Regional and Cross-Time Crop Mapping from Satellite Image Time Series},
  author = {Mohammadi, Sina and Belgiu, Mariana and Stein, Alfred},
  date = {2024-12-01},
  journaltitle = {Remote Sensing of Environment},
  shortjournal = {Remote Sensing of Environment},
  volume = {314},
  pages = {114385},
  issn = {0034-4257},
  doi = {10.1016/j.rse.2024.114385},
  url = {https://www.sciencedirect.com/science/article/pii/S0034425724004115},
  abstract = {Precise and timely information about crop types plays a crucial role in various agriculture-related applications. However, crop type mapping methods often face significant challenges in cross-regional and cross-time scenarios with high discrepancies between temporal-spectral characteristics of crops from different regions and years. Unsupervised domain adaptation (UDA) methods have been employed to mitigate the problem of domain shift between the source and target domains. Since these methods require source domain data during the adaptation phase, they demand significant computational resources and data storage, especially when large labeled crop mapping source datasets are available. This leads to increased energy consumption and financial costs. To address this limitation, we developed a source-free UDA method for cross-regional and cross-time crop mapping, capable of adapting the source-pretrained models to the target datasets without requiring the source datasets. The method mitigates the domain shift problem by leveraging mutual information loss. The diversity and discriminability terms in the loss function are balanced through a novel unsupervised weighting strategy based on mean confidence scores of the predicted categories. Our experiments on mapping corn, soybean, and the class Other from Landsat image time series in the U.S. demonstrated that the adapted models using different backbone networks outperformed their non-adapted counterparts. With CNN, Transformer, and LSTM backbone networks, our adaptation method increased the macro F1 scores by 12.9\%, 7.1\%, and 5.8\% on average in cross-time tests and by 20.1\%, 12.5\%, and 8.8\% on average in cross-regional tests, respectively. Additionally, in an experiment covering a large study area of 450 km × 300 km, the adapted model with the CNN backbone network obtained a macro F1 score of 92.6\%, outperforming its non-adapted counterpart with a macro F1 score of 89.2\%. Our experiments on mapping the same classes using Sentinel-2 image times series in France demonstrated the effectiveness of our method across different countries and sensors. We also tested our method in more diverse agricultural areas in Denmark and France containing six classes. The results showed that the adapted models outperformed the non-adapted models. Moreover, in within-season experiments, the adapted models performed better than the non-adapted models in the vast majority of weeks. These results and their comparison to those obtained by the other investigated UDA methods demonstrated the efficiency of our proposed method for both end-of-season and within-season crop mapping tasks. Additionally, our study showed that the method is modular and flexible in employing various backbone networks. The code and data are available at https://github.com/Sina-Mohammadi/SFUDA-CropMapping.},
  keywords = {Crop classification,Deep learning,In-season crop type mapping,Model generalization,Multi-temporal,Source-free unsupervised domain adaptation,Transfer learning},
  file = {C\:\\Users\\sjelic\\Zotero\\storage\\X75IUGWK\\Mohammadi et al. - 2024 - A source-free unsupervised domain adaptation method for cross-regional and cross-time crop mapping f.pdf;C\:\\Users\\sjelic\\Zotero\\storage\\HJPENYZW\\S0034425724004115.html}
}

@incollection{mondainiSteinerTreeProblem2008,
  title = {The {{Steiner Tree Problem}} and {{Its Application}} to the {{Modelling}} of {{Biomolecular Structures}}},
  booktitle = {Mathematical {{Modelling}} of {{Biosystems}}},
  author = {Mondaini, Rubem P.},
  editor = {Mondaini, Rubem P. and Pardalos, Panos M.},
  date = {2008},
  series = {Applied {{Optimization}}},
  pages = {199--219},
  publisher = {Springer},
  location = {Berlin, Heidelberg},
  doi = {10.1007/978-3-540-76784-8_6},
  url = {https://doi.org/10.1007/978-3-540-76784-8_6},
  abstract = {There is now a paradigm in the study of an efficient modelling of biomolecular structure. The foundations of the physics and the elucidation of biological function of living organisms will be asserted on a clear geometrical language according the best ideas of D’arcy Thompson [1], Rashevsky [2], Schrödinger [3] and Anfmsen [4]. The present work aims to give a possible mathematical description of one of Nature’s services of noticeable importance in the organization of life and its maintenance: the specific geometric form of macromolecular structure as provided by the mathematical problem of organization of Steiner Minimal Trees. The energy minimization process which lead to the formation of a biomacromolecule can be understood and modelled by the search process of organization of Steiner trees as the representatives of the possible molecular configurations corresponding to local minima of the free energy. Life maintenance and the survival of the living organism is guaranteed by the competence of staying away from the Global minimum structure and its associated Steiner Minimal Tree.},
  isbn = {978-3-540-76784-8},
  langid = {english},
  keywords = {Chebyshev polynomials,evenly spaced points,Fermat-Steiner problem,Steiner minimal tree,Steiner Ratio function}
}

@article{myungComparisonGroupSteiner2011,
  title = {A {{Comparison}} of {{Group Steiner Tree Formulations}}},
  author = {Myung, Young-Soo},
  date = {2011-09-01},
  journaltitle = {Journal of Korean Institute of Industrial Engineers},
  shortjournal = {Journal of Korean Institute of Industrial Engineers},
  volume = {37},
  doi = {10.7232/JKIIE.2011.37.3.191},
  abstract = {The group Steiner tree problem is a generalization of the Steiner tree problem that is defined as follows. Given a weighted graph with a family of subsets of nodes, called groups, the problem is to find a minimum weighted tree that contains at least one node in each group. We present some existing and some new formulations for the problem and compare the relaxations of such formulations.},
  file = {C:\Users\sjelic\Zotero\storage\24SPN3QK\Myung - 2011 - A Comparison of Group Steiner Tree Formulations.pdf}
}

@article{neves-moreiraTimeWindowAssignment2018,
  title = {The Time Window Assignment Vehicle Routing Problem with Product Dependent Deliveries},
  author = {Neves-Moreira, Fábio and Pereira da Silva, Diogo and Guimarães, Luís and Amorim, Pedro and Almada-Lobo, Bernardo},
  date = {2018-08-01},
  journaltitle = {Transportation Research Part E: Logistics and Transportation Review},
  shortjournal = {Transportation Research Part E: Logistics and Transportation Review},
  volume = {116},
  pages = {163--183},
  issn = {1366-5545},
  doi = {10.1016/j.tre.2018.03.004},
  url = {https://www.sciencedirect.com/science/article/pii/S1366554517310566},
  abstract = {This paper presents a new formulation for a time window assignment vehicle routing problem where time windows are defined for multiple product segments. This two-stage stochastic optimization problem is solved by means of a fix-and-optimize based matheuristic. The first stage assigns product dependent time windows while the second stage defines delivery schedules. Our approach outperforms a general-purpose solver and achieves an average cost decrease of 5.3\% over expected value problem approaches. Furthermore, a sensitivity analysis on three operational models shows that it is possible to obtain significant savings compared to the solutions provided by a large European food retailer.},
  keywords = {Fix-and-optimize,Multiple product deliveries,Retail operations,Stochastic optimization,Time window assignment,Vehicle routing},
  file = {C:\Users\sjelic\Zotero\storage\M5PHWJID\S1366554517310566.html}
}

@article{ngConvolutionalNeuralNetwork2019,
  title = {Convolutional Neural Network for Simultaneous Prediction of Several Soil Properties Using Visible/near-Infrared, Mid-Infrared, and Their Combined Spectra},
  author = {Ng, Wartini and Minasny, Budiman and Montazerolghaem, Maryam and Padarian, Jose and Ferguson, Richard and Bailey, Scarlett and McBratney, Alex B.},
  date = {2019-10-15},
  journaltitle = {Geoderma},
  shortjournal = {Geoderma},
  volume = {352},
  pages = {251--267},
  issn = {0016-7061},
  doi = {10.1016/j.geoderma.2019.06.016},
  url = {https://www.sciencedirect.com/science/article/pii/S0016706119300588},
  abstract = {No single instrument can characterize all soil properties because soil is a complex material. With the advancement of technology, laboratories have become equipped with various spectrometers. By fusing output from different spectrometers, better prediction outcomes are expected than using any single spectrometer alone. In this study, model performance from a single spectrometer (visible-near-infrared spectroscopy, vis-NIR or mid-infrared spectroscopy, MIR) was compared to the combined spectrometers (vis-NIR and MIR). We selected a total of 14,594 samples from the Kellogg Soil Survey Laboratory (KSSL) database that had both vis-NIR and MIR spectra along with measurements of sand, clay, total C (TC) content, organic C (OC) content, cation exchange capacity (CEC), and pH. The dataset was randomly split into 75\% training (n\,=\,10,946) and the remaining (n\,=\,3,648) as a test set. Prediction models were constructed with partial least squares regression (PLSR) and Cubist tree model. Additionally, we explored the use of a deep learning model, the convolutional neural network (CNN). We investigated various ways to feed spectral data to the CNN, either as one-dimensional (1D) data (as a spectrum) or as two-dimensional (2D) data (as a spectrogram). Compared to the PLSR model, we found that the CNN model provides an average improvement prediction of 33–42\% using vis-NIR and 30–43\% using MIR spectral data input. The relative accuracy improvement of CNN, when compared to the Cubist regression tree model, ranged between 22 and 36\% with vis-NIR and 16–27\% with MIR spectral data input. Various methods to fuse the vis-NIR and MIR spectral data were explored. We compared the performance of spectral concatenation (for PLSR and Cubist model), two-channel input method, and outer product analysis (OPA) method (for CNN model). We found that the performance of two-channel 1D CNN model was the best (R2\,=\,0.95–0.98) followed closely by the OPA with CNN (R2\,=\,0.93–0.98), Cubist model with spectral concatenation (R2\,=\,0.91–0.97), two-channel 2D CNN model (R2\,=\,0.90–0.95) and PLSR with spectral concatenation (R2\,=\,0.87–0.95). Chemometric analysis of spectroscopy data relied on spectral pre-processing methods: such as spectral trimming, baseline correction, smoothing, and normalization before being fed into the model. CNN achieved higher performance than the PLSR and Cubist model without utilizing the pre-processed spectral data. We also found that the predictions using the CNN model retained similar correlations to the actual values in comparison to other models. By doing sensitivity analysis, we identified the important spectral wavelengths variables used by the CNN model to predict various soil properties. CNN is an effective model for modelling soil properties from a large spectral library.},
  keywords = {Deep learning,Mid-infrared spectroscopy,Multi-task learning,Near-infrared spectroscopy},
  file = {C\:\\Users\\sjelic\\Zotero\\storage\\76ZBM8VR\\Ng и сар. - 2019 - Convolutional neural network for simultaneous prediction of several soil properties using visiblene.pdf;C\:\\Users\\sjelic\\Zotero\\storage\\QHUCGQPT\\S0016706119300588.html}
}

@article{ngMidinfraredSpectroscopyAccurate2022,
  title = {Mid-Infrared Spectroscopy for Accurate Measurement of an Extensive Set of Soil Properties for Assessing Soil Functions},
  author = {Ng, Wartini and Minasny, Budiman and Jeon, Sang Ho and McBratney, Alex},
  date = {2022-03-01},
  journaltitle = {Soil Security},
  shortjournal = {Soil Security},
  volume = {6},
  pages = {100043},
  issn = {2667-0062},
  doi = {10.1016/j.soisec.2022.100043},
  url = {https://www.sciencedirect.com/science/article/pii/S2667006222000107},
  abstract = {Quantitative assessment of soil functions requires the characterization of soil capability and condition. Mid-infrared (MIR) spectroscopy has been suggested as a viable alternative to the wet chemistry method. However, the extensive set of soil properties that can be well predicted have yet to be explored. The USDA MIR spectral library contains approximately 45,000 samples with more than 119 soil properties. This unique dataset allows us to establish which soil properties that can be accurately measured. Memory-based learning (MBL) algorithm achieved higher accuracy than the Cubist. The prediction accuracies of different properties were then categorized further into four classes (A, B, C, and D) based on four accuracy metrics, namely coefficient of determination (R2), Lin's concordance correlation coefficient (LCCC), ratio of performance to the interquartile range (RPIQ), and standardized bias (Stb). We found that a single MIR spectrum can infer 50 soil properties with high accuracy (A or B category), and 44 properties can be estimated approximately (C category). These properties can then be used to evaluate a range of soil functions, including food production, carbon storage, water storage, nutrient cycling, and habitat function.},
  keywords = {Chemometrics,Cubist,Memory-based learning,Mid-infrared spectroscopy,Soil spectral library},
  file = {C\:\\Users\\sjelic\\Zotero\\storage\\BFA3XBID\\Ng et al. - 2022 - Mid-infrared spectroscopy for accurate measurement.pdf;C\:\\Users\\sjelic\\Zotero\\storage\\CEDGT6PA\\S2667006222000107.html}
}

@article{nguyenFastApproximationAlgorithm2014,
  title = {A Fast Approximation Algorithm for Solving the Complete Set Packing Problem},
  author = {Nguyen, Tri-Dung},
  date = {2014-08-16},
  journaltitle = {European Journal of Operational Research},
  shortjournal = {European Journal of Operational Research},
  volume = {237},
  number = {1},
  pages = {62--70},
  issn = {0377-2217},
  doi = {10.1016/j.ejor.2014.01.024},
  url = {https://www.sciencedirect.com/science/article/pii/S0377221714000459},
  abstract = {We study the complete set packing problem (CSPP) where the family of feasible subsets may include all possible combinations of objects. This setting arises in applications such as combinatorial auctions (for selecting optimal bids) and cooperative game theory (for finding optimal coalition structures). Although the set packing problem has been well-studied in the literature, where exact and approximation algorithms can solve very large instances with up to hundreds of objects and thousands of feasible subsets, these methods are not extendable to the CSPP since the number of feasible subsets is exponentially large. Formulating the CSPP as an MILP and solving it directly, using CPLEX for example, is impossible for problems with more than 20 objects. We propose a new mathematical formulation for the CSPP that directly leads to an efficient algorithm for finding feasible set packings (upper bounds). We also propose a new formulation for finding tighter lower bounds compared to LP relaxation and develop an efficient method for solving the corresponding large-scale MILP. We test the algorithm with the winner determination problem in spectrum auctions, the coalition structure generation problem in coalitional skill games, and a number of other simulated problems that appear in the literature.},
  keywords = {Coalition structure generation,Combinatorial auctions,Large-scale MILP,Set packing problem,Winner determination problem},
  file = {C\:\\Users\\sjelic\\Zotero\\storage\\XE3L32AP\\Nguyen - 2014 - A fast approximation algorithm for solving the com.pdf;C\:\\Users\\sjelic\\Zotero\\storage\\EUR8IXWZ\\S0377221714000459.html}
}

@article{nowakowskiCropTypeMapping2021,
  title = {Crop Type Mapping by Using Transfer Learning},
  author = {Nowakowski, Artur and Mrziglod, John and Spiller, Dario and Bonifacio, Rogerio and Ferrari, Irene and Mathieu, Pierre Philippe and Garcia-Herranz, Manuel and Kim, Do-Hyung},
  date = {2021-06},
  journaltitle = {International Journal of Applied Earth Observation and Geoinformation},
  shortjournal = {International Journal of Applied Earth Observation and Geoinformation},
  volume = {98},
  pages = {102313},
  issn = {15698432},
  doi = {10.1016/j.jag.2021.102313},
  url = {https://linkinghub.elsevier.com/retrieve/pii/S0303243421000209},
  abstract = {Crop type mapping currently represents an important problem in remote sensing. Accurate information on the extent and types of crops derived from remote sensing can help managing and improving agriculture especially for developing countries where such information is scarce. In this paper, high-resolution RGB drone images are the input data for the classification performed using a transfer learning (TL) approach. VGG16 and GoogLeNet, which are pre-trained convolutional neural networks (CNNs) used for classification tasks coming from computer vision, are considered for the mapping of the crop types. Thanks to the transferred knowledge, the proposed models can successfully classify the studied crop types with high overall accuracy for two considered cases, achieving up to almost 83\% for the Malawi dataset and up to 90\% for the Mozambique dataset. Notably, these results are comparable to the ones achieved by the same deep CNN architectures in many computer vision tasks. With regard to drone data analysis, application of deep CNN is very limited so far due to high requirements on the number of samples needed to train such complicated architectures. Our results demonstrate that the transfer learning is an efficient way to overcome this problem and take full advantage of the benefits of deep CNN ar­ chitectures for drone-based crop type mapping. Moreover, based on experiments with different TL approaches we show that the number of frozen layers is an important parameter of TL and a fine-tuning of all the CNN weights results in significantly better performance than the approaches that apply fine-tuning only on some numbers of last layers.},
  langid = {english},
  file = {C:\Users\sjelic\Zotero\storage\2DM99S7H\Nowakowski et al. - 2021 - Crop type mapping by using transfer learning.pdf}
}

@article{panSurveyTransferLearning2010,
  title = {A {{Survey}} on {{Transfer Learning}}},
  author = {Pan, Sinno Jialin and Yang, Qiang},
  date = {2010-10},
  journaltitle = {IEEE Transactions on Knowledge and Data Engineering},
  shortjournal = {IEEE Trans. Knowl. Data Eng.},
  volume = {22},
  number = {10},
  pages = {1345--1359},
  issn = {1041-4347},
  doi = {10.1109/TKDE.2009.191},
  url = {http://ieeexplore.ieee.org/document/5288526/},
  abstract = {A major assumption in many machine learning and data mining algorithms is that the training and future data must be in the same feature space and have the same distribution. However, in many real-world applications, this assumption may not hold. For example, we sometimes have a classification task in one domain of interest, but we only have sufficient training data in another domain of interest, where the latter data may be in a different feature space or follow a different data distribution. In such cases, knowledge transfer, if done successfully, would greatly improve the performance of learning by avoiding much expensive data-labeling efforts. In recent years, transfer learning has emerged as a new learning framework to address this problem. This survey focuses on categorizing and reviewing the current progress on transfer learning for classification, regression, and clustering problems. In this survey, we discuss the relationship between transfer learning and other related machine learning techniques such as domain adaptation, multitask learning and sample selection bias, as well as covariate shift. We also explore some potential future issues in transfer learning research.},
  langid = {english},
  file = {C:\Users\sjelic\Zotero\storage\LH25FMY6\Pan and Yang - 2010 - A Survey on Transfer Learning.pdf}
}

@article{pargentRegularizedTargetEncoding2022,
  title = {Regularized Target Encoding Outperforms Traditional Methods in Supervised Machine Learning with High Cardinality Features},
  author = {Pargent, Florian and Pfisterer, Florian and Thomas, Janek and Bischl, Bernd},
  date = {2022-11-01},
  journaltitle = {Computational Statistics},
  shortjournal = {Comput Stat},
  volume = {37},
  number = {5},
  pages = {2671--2692},
  issn = {1613-9658},
  doi = {10.1007/s00180-022-01207-6},
  url = {https://doi.org/10.1007/s00180-022-01207-6},
  abstract = {Since most machine learning (ML) algorithms are designed for numerical inputs, efficiently encoding categorical variables is a crucial aspect in data analysis. A common problem are high cardinality features, i.e.~unordered categorical predictor variables with a high number of levels. We study techniques that yield numeric representations of categorical variables which can then be used in subsequent ML applications. We focus on the impact of these techniques on a subsequent algorithm’s predictive performance, and—if possible—derive best practices on when to use which technique. We conducted a large-scale benchmark experiment, where we compared different encoding strategies together with five ML algorithms (lasso, random forest, gradient boosting, k-nearest neighbors, support vector machine) using datasets from regression, binary- and multiclass–classification settings. In our study, regularized versions of target encoding (i.e.~using target predictions based on the feature levels in the training set as a new numerical feature) consistently provided the best results. Traditionally widely used encodings that make unreasonable assumptions to map levels to integers (e.g.~integer encoding) or to reduce the number of levels (possibly based on target information, e.g.~leaf encoding) before creating binary indicator variables (one-hot or dummy encoding) were not as effective in comparison.},
  langid = {english},
  keywords = {Benchmark,Dummy encoding,Generalized linear mixed models,High-cardinality categorical features,Supervised machine learning,Target encoding},
  file = {C:\Users\sjelic\Zotero\storage\JF8A5LFL\Pargent et al. - 2022 - Regularized target encoding outperforms traditional methods in supervised machine learning with high.pdf}
}

@article{parkAlgorithmsGeneralizedClusterwise2017,
  title = {Algorithms for {{Generalized Clusterwise Linear Regression}}},
  author = {Park, Young Woong and Jiang, Yan and Klabjan, Diego and Williams, Loren},
  date = {2017-04-05},
  journaltitle = {INFORMS Journal on Computing},
  publisher = {INFORMS},
  doi = {10.1287/ijoc.2016.0729},
  url = {https://pubsonline.informs.org/doi/abs/10.1287/ijoc.2016.0729},
  abstract = {Clusterwise linear regression (CLR), a clustering problem intertwined with regression, finds clusters of entities such that the overall sum of squared errors from regressions performed over these c...},
  langid = {english},
  file = {C\:\\Users\\sjelic\\Zotero\\storage\\AAP7ENXN\\Park и сар. - 2017 - Algorithms for Generalized Clusterwise Linear Regr.pdf;C\:\\Users\\sjelic\\Zotero\\storage\\FYMXCY53\\ijoc.2016.html}
}

@article{pengDomainAdaptationRemote2022,
  title = {Domain {{Adaptation}} in {{Remote Sensing Image Classification}}: {{A Survey}}},
  shorttitle = {Domain {{Adaptation}} in {{Remote Sensing Image Classification}}},
  author = {Peng, Jiangtao and Huang, Yi and Sun, Weiwei and Chen, Na and Ning, Yujie and Du, Qian},
  date = {2022},
  journaltitle = {IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing},
  volume = {15},
  pages = {9842--9859},
  issn = {2151-1535},
  doi = {10.1109/JSTARS.2022.3220875},
  url = {https://ieeexplore.ieee.org/document/9944086},
  abstract = {Traditional remote sensing (RS) image classification methods heavily rely on labeled samples for model training. When labeled samples are unavailable or labeled samples have different distributions from that of the samples to be classified, the classification model may fail. The cross-domain or cross-scene remote sensing image classification is developed for this case where an existing image for training and an unknown image from different scenes or domains for classification. The distribution inconsistency problem may be caused by the differences in acquisition environment conditions, acquisition scene, acquisition time, and/or changing sensors. To cope with the cross-domain remote sensing image classification problem, many domain adaptation (DA) techniques have been developed. In this article, we review DA methods in the fields of RS, especially hyperspectral image classification, and provide a survey of DA methods into traditional shallow DA methods (e.g., instance-based, feature-based, and classifier-based adaptations) and recently developed deep DA methods (e.g., discrepancy-based and adversarial-based adaptations).},
  eventtitle = {{{IEEE Journal}} of {{Selected Topics}} in {{Applied Earth Observations}} and {{Remote Sensing}}},
  keywords = {Adaptation models,Cross-domain classification,distribution difference,domain adaptation (DA),Feature extraction,Hyperspectral imaging,Image sensors,Kernel,Principal component analysis,remote sensing (RS) image,Sun},
  file = {C\:\\Users\\sjelic\\Zotero\\storage\\H8SHNICE\\Peng et al. - 2022 - Domain Adaptation in Remote Sensing Image Classification A Survey.pdf;C\:\\Users\\sjelic\\Zotero\\storage\\JZVK254S\\9944086.html}
}

@article{pennApproximationAlgorithmGroup2007,
  title = {Approximation Algorithm for the Group {{Steiner}} Network Problem},
  author = {Penn, Michal and Rozenfeld, Stas},
  date = {2007-03-01},
  journaltitle = {Networks},
  shortjournal = {Netw.},
  volume = {49},
  number = {2},
  pages = {160--167},
  issn = {0028-3045},
  abstract = {In this article we study the group Steiner network problem, which is defined in the following way. Given a graph G = (V,E), a partition of its vertices into K groups and connectivity requirements between the different groups, the aim is to find simultaneously a set of representatives, one for each group, and a minimum cost connected subgraph that satisfies the connectivity requirements between the groups (representatives). This problem is a generalization of the Steiner network problem and the group Steiner tree problem, two known NP-complete problems. We present an approximation algorithm for a special case of the group Steiner network problem with an approximation ratio of min \{2(1 + 2x),2I\}, where I is the cardinality of the largest group and x is a parameter that depends on the cost function. © 2006 Wiley Periodicals, Inc. NETWORKS, Vol. 49(2), 160–167 2007},
  keywords = {approximation algorithm,group Steiner network,integer and linear programming},
  file = {C:\Users\sjelic\Downloads\net.20151.pdf}
}

@book{pinedoSchedulingTheoryAlgorithms2022,
  title = {Scheduling: {{Theory}}, {{Algorithms}}, and {{Systems}}},
  shorttitle = {Scheduling},
  author = {Pinedo, Michael L.},
  date = {2022},
  publisher = {Springer International Publishing},
  location = {Cham},
  doi = {10.1007/978-3-031-05921-6},
  url = {https://link.springer.com/10.1007/978-3-031-05921-6},
  isbn = {978-3-031-05920-9 978-3-031-05921-6},
  langid = {english},
  keywords = {Applied Optimization,Complexity Theory,Decision Support Systems,Mathematical Programming,Stochastic Modeling}
}

@article{pizarroMultipleComparisonProcedures2002,
  title = {Multiple Comparison Procedures Applied to Model Selection},
  author = {Pizarro, Joaquı́n and Guerrero, Elisa and Galindo, Pedro L.},
  date = {2002-10-01},
  journaltitle = {Neurocomputing},
  shortjournal = {Neurocomputing},
  volume = {48},
  number = {1},
  pages = {155--173},
  issn = {0925-2312},
  doi = {10.1016/S0925-2312(01)00653-1},
  url = {https://www.sciencedirect.com/science/article/pii/S0925231201006531},
  abstract = {This paper presents a new approach to model selection based on hypothesis testing. We first describe a procedure to generate different scores for any candidate model from a single sample of training data and then discuss how to apply multiple comparison procedures (MCP) to model selection. MCP statistical tests allow us to compare three or more groups of data while controlling the probability of making at least one Type I error. The complete procedure is illustrated on several model selection tasks, including the determination of the number of hidden units for feed-forward neural networks and the number of kernels for RBF networks.},
  langid = {english},
  keywords = {Generalization,Model selection,Multiple comparison procedures,Network size,Problem complexity},
  file = {C:\Users\sjelic\Zotero\storage\8RW73KDU\Pizarro и сар. - 2002 - Multiple comparison procedures applied to model se.pdf}
}

@inproceedings{plaiaConstrainedClusterwiseLinear2005,
  title = {Constrained {{Clusterwise Linear Regression}}},
  booktitle = {New {{Developments}} in {{Classification}} and {{Data Analysis}}},
  author = {Plaia, Antonella},
  editor = {Bock, H.-H. and Gaul, W. and Vichi, M. and family=Arabie, given=Ph., given-i={{Ph}} and Baier, D. and Critchley, F. and Decker, R. and Diday, E. and Greenacre, M. and Lauro, C. and Meulman, J. and Monari, P. and Nishisato, S. and Ohsumi, N. and Opitz, O. and Ritter, G. and Schader, M. and Weihs, C. and Vichi, Maurizio and Monari, Paola and Mignani, Stefania and Montanari, Angela},
  date = {2005},
  series = {Studies in {{Classification}}, {{Data Analysis}}, and {{Knowledge Organization}}},
  pages = {79--86},
  publisher = {Springer},
  location = {Berlin, Heidelberg},
  doi = {10.1007/3-540-27373-5_10},
  abstract = {In market segmentation, Conjoint Analysis is often used to estimate the importance of a product attributes at the level of each single customer, clustering, successively, the customers whose behavior can be considered similar. The preference model parameter estimation is made considering data (usually opinions) of a single customer at a time, but these data are usually very few as each customer is called to express his opinion about a small number of different products (in order to simplify his/her work). In the present paper a Constrained Clusterwise Linear Regression algorithm is presented, that allows simultaneously to estimate parameters and to cluster customers, using, for the estimation, the data of all the customers with similar behavior.},
  isbn = {978-3-540-27373-8},
  langid = {english},
  keywords = {Attribute Level,Conjoint Analysis,Importance Weight,Ordinary Little Square,Preference Model},
  file = {C:\Users\sjelic\Zotero\storage\IJ57LTXE\Plaia - 2005 - Constrained Clusterwise Linear Regression.pdf}
}

@article{popGeneralizedMinimumSpanning2020,
  title = {The Generalized Minimum Spanning Tree Problem: {{An}} Overview of Formulations, Solution Procedures and Latest Advances},
  shorttitle = {The Generalized Minimum Spanning Tree Problem},
  author = {Pop, Petrică C.},
  date = {2020-05-16},
  journaltitle = {European Journal of Operational Research},
  shortjournal = {European Journal of Operational Research},
  volume = {283},
  number = {1},
  pages = {1--15},
  issn = {0377-2217},
  doi = {10.1016/j.ejor.2019.05.017},
  url = {https://www.sciencedirect.com/science/article/pii/S0377221719304217},
  abstract = {In this paper, some of the main known results relative to the generalized minimum spanning tree problem are surveyed. The principal feature of this problem is related to the fact that the vertices of the graph are partitioned into a certain number of clusters and we are interested in finding a minimum-cost tree spanning a subset of vertices with precisely one vertex considered from every cluster. The paper is structured around the following main headings: problem definition, variations and practical applications, complexity aspects, integer programming formulations, exact and heuristic solution approaches developed for solving this problem. Furthermore, we also discuss some open problems and possible research directions.},
  keywords = {Combinatorial optimization,Generalized minimum spanning tree problem,Survey},
  file = {C:\Users\sjelic\Zotero\storage\FFJRVUNM\S0377221719304217.html}
}

@online{Proceedings2015Annual,
  title = {Proceedings of the 2015 {{Annual ACM-SIAM Symposium}} on {{Discrete Algorithms}} ({{SODA}}) | {{On Survivable Set Connectivity}}},
  url = {https://epubs.siam.org/doi/abs/10.1137/1.9781611973730.3},
  file = {C:\Users\sjelic\Zotero\storage\W2HP5D38\1.9781611973730.html}
}

@incollection{ramalingamCloudHostedEnsemble2022,
  title = {Cloud {{Hosted Ensemble Learning-Based Rental Apartment Price Prediction Model Using Stacking Technique}}},
  author = {Ramalingam, Rajasekar and Soundrapandiyan, Rajkumar},
  date = {2022-11-25},
  pages = {229--238},
  doi = {10.4018/978-1-6684-6001-6.ch015},
  abstract = {In this chapter, online rental listings of the city of Hyderabad are used as a data source for mapping house rent. Data points were scraped from one of the popular Indian rental websites www.nobroker.in. With the collected information, models of rental market dynamics were developed and evaluated using regression and boosting algorithms such as AdaBoost, CatBoost, LightGBM, XGBoost, KRR, ENet, and Lasso regression. An ensemble machine learning algorithm of the best combination of the aforementioned algorithms was also implemented using the stacking technique. The results of these algorithms were compared using several performance metrics such as coefficient of determination (R2 score), mean squared error (MSE), root mean squared error (RMSE), mean absolute error (MAE), and accuracy in order to determine the most effective model. According to further examination of results, it is clear that the ensemble machine learning algorithm does outperform the others in terms of better accuracy and reduced errors.},
  isbn = {978-1-6684-6001-6 978-1-6684-6003-0 978-1-6684-6002-3},
  file = {C:\Users\sjelic\Zotero\storage\PN6ZUPRS\Ramalingam and Soundrapandiyan - 2022 - Cloud Hosted Ensemble Learning-Based Rental Apartm.pdf}
}

@article{reedMeasuringPhenologicalVariability1994,
  title = {Measuring Phenological Variability from Satellite Imagery},
  author = {Reed, Bradley C. and Brown, Jesslyn F. and VanderZee, Darrel and Loveland, Thomas R. and Merchant, James W. and Ohlen, Donald O.},
  date = {1994},
  journaltitle = {Journal of Vegetation Science},
  volume = {5},
  number = {5},
  pages = {703--714},
  issn = {1654-1103},
  doi = {10.2307/3235884},
  url = {https://onlinelibrary.wiley.com/doi/abs/10.2307/3235884},
  abstract = {Abstract. Vegetation phenological phenomena are closely related to seasonal dynamics of the lower atmosphere and are therefore important elements in global models and vegetation monitoring. Normalized difference vegetation index (NDVI) data derived from the National Oceanic and Atmospheric Administration's Advanced Very High Resolution Radiometer (AVHRR) satellite sensor offer a means of efficiently and objectively evaluating phenological characteristics over large areas. Twelve metrics linked to key phenological events were computed based on time-series NDVI data collected from 1989 to 1992 over the conterminous United States. These measures include the onset of greenness, time of peak NDVI, maximum NDVI, rate of greenup, rate of senescence, and integrated NDVI. Measures of central tendency and variability of the measures were computed and analyzed for various land cover types. Results from the analysis showed strong coincidence between the satellite-derived metrics and predicted phenological characteristics. In particular, the metrics identified interannual variability of spring wheat in North Dakota, characterized the phenology of four types of grasslands, and established the phenological consistency of deciduous and coniferous forests. These results have implications for large-area land cover mapping and monitoring. The utility of remotely sensed data as input to vegetation mapping is demonstrated by showing the distinct phenology of several land cover types. More stable information contained in ancillary data should be incorporated into the mapping process, particularly in areas with high phenological variability. In a regional or global monitoring system, an increase in variability in a region may serve as a signal to perform more detailed land cover analysis with higher resolution imagery.},
  langid = {english},
  keywords = {GIS,Land cover,Remote sensing,Time-series analysis,Vegetation monitoring},
  file = {C:\Users\sjelic\Zotero\storage\L5P2VG7S\3235884.html}
}

@article{rehfeldtExactSolutionPrizeCollecting2022,
  title = {On the {{Exact Solution}} of {{Prize-Collecting Steiner Tree Problems}}},
  author = {Rehfeldt, Daniel and Koch, Thorsten},
  date = {2022-03},
  journaltitle = {INFORMS Journal on Computing},
  volume = {34},
  number = {2},
  pages = {872--889},
  publisher = {INFORMS},
  issn = {1091-9856},
  doi = {10.1287/ijoc.2021.1087},
  url = {https://pubsonline.informs.org/doi/abs/10.1287/ijoc.2021.1087},
  abstract = {The prize-collecting Steiner tree problem (PCSTP) is a well-known generalization of the classic Steiner tree problem in graphs, with a large number of practical applications. It attracted particular interest during the 11th DIMACS Challenge in 2014, and since then, several PCSTP solvers have been introduced in the literature. Although these new solvers further, and often drastically, improved on the results of the DIMACS Challenge, many PCSTP benchmark instances have remained unsolved. The following article describes further advances in the state of the art in exact PCSTP solving. It introduces new techniques and algorithms for PCSTP, involving various new transformations (or reductions) of PCSTP instances to equivalent problems, for example, to decrease the problem size or to obtain a better integer programming formulation. Several of the new techniques and algorithms provably dominate previous approaches. Further theoretical properties of the new components, such as their complexity, are discussed. Also, new complexity results for the exact solution of PCSTP and related problems are described, which form the base of the algorithm design. Finally, the new developments also translate into a strong computational performance: the resulting exact PCSTP solver outperforms all previous approaches, both in terms of runtime and solvability. In particular, it solves several formerly intractable benchmark instances from the 11th DIMACS Challenge to optimality. Moreover, several recently introduced large-scale instances with up to 10 million edges, previously considered to be too large for any exact approach, can now be solved to optimality in less than two hours. Summary of Contribution: The prize-collecting Steiner tree problem (PCSTP) is a well-known generalization of the classic Steiner tree problem in graphs, with many practical applications. The article introduces and analyses new techniques and algorithms for PCSTP that ultimately aim for improved (practical) exact solution. The algorithmic developments are underpinned by results on theoretical aspects, such as fixed-parameter tractability of PCSTP. Computationally, we considerably push the limits of tractibility, being able to solve PCSTP instances with up to 10 million edges. The new solver, which also considerably outperforms the state of the art on smaller instances, will be made publicly available as part of the SCIP Optimization Suite.},
  keywords = {complexity,exact solution,maximum-weight connected subgraph problem,node-weighted Steiner tree,prize-collecting Steiner tree},
  file = {C:\Users\sjelic\Downloads\pcstpZIBv2.pdf}
}

@inproceedings{reichSteinerProblemVLSI1990,
  title = {Beyond {{Steiner}}'s Problem: {{A VLSI}} Oriented Generalization},
  shorttitle = {Beyond {{Steiner}}'s Problem},
  booktitle = {Graph-{{Theoretic Concepts}} in {{Computer Science}}},
  author = {Reich, Gabriele and Widmayer, Peter},
  editor = {Nagl, Manfred},
  date = {1990},
  series = {Lecture {{Notes}} in {{Computer Science}}},
  pages = {196--210},
  publisher = {Springer},
  location = {Berlin, Heidelberg},
  doi = {10.1007/3-540-52292-1_14},
  abstract = {We consider a generalized version of Steiner's problem in graphs, motivated by the wire routing phase in physical VLSI design: given a connected, undirected distance graph with groups of required vertices and Steiner vertices, find a shortest connected subgraph containing at least one required vertex of each group. We propose two efficient approximation algorithms computing different approximate solutions in time O(|E| + |V|log|V|) and in time O(g · (|E| + |V|log|V|)), respectively, where |E| is the number of edges in the given graph, |V| is the number of vertices, and g is the number of groups. The latter algorithm propagates a set of wavefronts with different distances simultaneously through the graph; it is interesting in its own right.},
  isbn = {978-3-540-46950-6},
  langid = {english},
  keywords = {Approximation Algorithm,Priority Queue,Steiner Minimal Tree,Steiner Problem,Steiner Tree}
}

@article{renEnsembleSurrogatesCombining2022,
  title = {Ensemble of Surrogates Combining {{Kriging}} and {{Artificial Neural Networks}} for Reliability Analysis with Local Goodness Measurement},
  author = {Ren, Chao and Aoues, Younes and Lemosse, Didier and Souza De Cursi, Eduardo},
  date = {2022-05-01},
  journaltitle = {Structural Safety},
  shortjournal = {Structural Safety},
  volume = {96},
  pages = {102186},
  issn = {0167-4730},
  doi = {10.1016/j.strusafe.2022.102186},
  url = {https://www.sciencedirect.com/science/article/pii/S0167473022000029},
  abstract = {In engineering problems, reliability assessment is often computationally expensive. Surrogate models with active learning approaches are commonly used to solve this problem. Kriging is one of the most popular surrogate models used in this domain. Recently, artificial neural networks (ANN) have also attracted a lot of interest for structural reliability assessment due to their powerful capability. Many active learning approaches have been developed based on Kriging models and ANN. But selecting an appropriate model or technique for a reliability assessment problem with limited knowledge of the limit state function remains a challenging task. Ensemble of surrogates seems to be a good approach to tackle this challenge. In this work, two active learning approaches are proposed to combine Kriging and ANN models for reliability analysis. One is the local best surrogate (LBS) approach and the other is the local weighted average surrogate (LWAS) approach. Cross-validation and Jackknife techniques are used to estimate prediction errors of the surrogate models. In addition, two methods are proposed to locally measure the goodness of the surrogate models and calculate the prediction errors of the ensemble of surrogates. The surrogate models are updated by selecting the new sample points that have large prediction errors and are close to the limit state. The efficiency and accuracy of the proposed approaches are demonstrated by 4 representative examples and two finite element problems. The results show that the proposed methods can be effective in evaluating the reliability of high dimension and rare event problems with less computational costs than the single typical surrogate model with active learning approaches (e.g. AK-MCS). Moreover, compared with the ensemble of surrogate models based on global goodness measurement, the proposed approaches also outperform in most cases. Finally, it should be mentioned that the proposed approaches are not only suitable for the combination of Kriging and ANN but also can be extended to other multiple surrogate models including support vector machine, polynomial chaos expansion, and so on.},
  keywords = {Active learning,Artificial Neural Networks,Ensemble of surrogate models,Kriging,Reliability analysis},
  file = {C:\Users\sjelic\Zotero\storage\Q57W6SD9\S0167473022000029.html}
}

@online{RobustLinearRegression,
  title = {Robust Linear Regression for High‐dimensional Data: {{An}} Overview - {{Filzmoser}} - 2021 - {{WIREs Computational Statistics}} - {{Wiley Online Library}}},
  url = {https://wires.onlinelibrary.wiley.com/doi/full/10.1002/wics.1524},
  file = {C\:\\Users\\sjelic\\Zotero\\storage\\BGVSIK4J\\Robust linear regression for high‐dimensional data.pdf;C\:\\Users\\sjelic\\Zotero\\storage\\WC8C9HX2\\wics.html}
}

@online{rothvossDirectedSteinerTree2012,
  title = {Directed {{Steiner Tree}} and the {{Lasserre Hierarchy}}},
  author = {Rothvoß, Thomas},
  date = {2012-06-11},
  eprint = {1111.5473},
  eprinttype = {arXiv},
  eprintclass = {cs},
  url = {http://arxiv.org/abs/1111.5473},
  abstract = {The goal for the Directed Steiner Tree problem is to find a minimum cost tree in a directed graph G=(V,E) that connects all terminals X to a given root r. It is well known that modulo a logarithmic factor it suffices to consider acyclic graphs where the nodes are arranged in L {$<$}= log |X| levels. Unfortunately the natural LP formulation has a |X|\textasciicircum (1/2) integrality gap already for 5 levels. We show that for every L, the O(L)-round Lasserre Strengthening of this LP has integrality gap O(L log |X|). This provides a polynomial time |X|\textasciicircum\{epsilon\}-approximation and a O(log\textasciicircum 3 |X|) approximation in O(n\textasciicircum\{log |X|) time, matching the best known approximation guarantee obtained by a greedy algorithm of Charikar et al.\vphantom\}},
  pubstate = {prepublished},
  keywords = {Computer Science - Data Structures and Algorithms,F.2.0},
  file = {C\:\\Users\\sjelic\\Zotero\\storage\\CUXM4ZGQ\\Rothvoß - 2012 - Directed Steiner Tree and the Lasserre Hierarchy.pdf;C\:\\Users\\sjelic\\Zotero\\storage\\ZLPRHGF8\\1111.html}
}

@inproceedings{russwurmBREIZHCROPSTIMESERIES2020,
  title = {{{BREIZHCROPS}}: {{A TIME SERIES DATASET FOR CROP TYPE MAPPING}}},
  shorttitle = {{{BREIZHCROPS}}},
  booktitle = {The {{International Archives}} of the {{Photogrammetry}}, {{Remote Sensing}} and {{Spatial Information Sciences}}},
  author = {Rußwurm, M. and Pelletier, C. and Zollner, M. and Lefèvre, S. and Körner, M.},
  date = {2020-08-14},
  volume = {XLIII-B2-2020},
  pages = {1545--1551},
  publisher = {Copernicus GmbH},
  issn = {1682-1750},
  doi = {10.5194/isprs-archives-XLIII-B2-2020-1545-2020},
  url = {https://www.int-arch-photogramm-remote-sens-spatial-inf-sci.net/XLIII-B2-2020/1545/2020/},
  abstract = {{$<$}p{$><$}strong class="journal-contentHeaderColor"{$>$}Abstract.{$<$}/strong{$>$} We present BreizhCrops, a novel benchmark dataset for the supervised classification of field crops from satellite time series. We aggregated label data and Sentinel-2 top-of-atmosphere as well as bottom-of-atmosphere time series in the region of Brittany (Breizh in local language), north-east France. We compare seven recently proposed deep neural networks along with a Random Forest baseline. The dataset, model (re-)implementations and pre-trained model weights are available at the associated GitHub repository (https://github.com/dl4sits/breizhcrops) that has been designed with applicability for practitioners in mind. We plan to maintain the repository with additional data and welcome contributions of novel methods to build a state-of-the-art benchmark on methods for crop type mapping.{$<$}/p{$>$}},
  eventtitle = {{{XXIV ISPRS Congress}}, {{Commission II}} ({{Volume XLIII-B2-2020}}) - 2020 Edition},
  langid = {english},
  file = {C\:\\Users\\sjelic\\Zotero\\storage\\8NDUJ5PU\\Rußwurm et al. - 2020 - BREIZHCROPS A TIME SERIES DATASET FOR CROP TYPE M.pdf;C\:\\Users\\sjelic\\Zotero\\storage\\LN8UGXZ3\\2020.html}
}

@article{russwurmSelfattentionRawOptical2020,
  title = {Self-Attention for Raw Optical {{Satellite Time Series Classification}}},
  author = {Rußwurm, Marc and Körner, Marco},
  date = {2020-11-01},
  journaltitle = {ISPRS Journal of Photogrammetry and Remote Sensing},
  shortjournal = {ISPRS Journal of Photogrammetry and Remote Sensing},
  volume = {169},
  pages = {421--435},
  issn = {0924-2716},
  doi = {10.1016/j.isprsjprs.2020.06.006},
  url = {https://www.sciencedirect.com/science/article/pii/S0924271620301647},
  abstract = {The amount of available Earth observation data has increased dramatically in recent years. Efficiently making use of the entire body of information is a current challenge in remote sensing; it demands lightweight problem-agnostic models that do not require region- or problem-specific expert knowledge. End-to-end trained deep learning models can make use of raw sensory data by learning feature extraction and classification in one step, solely from data. Still, many methods proposed in remote sensing research require implicit feature extraction through data preprocessing or explicit design of features. In this work, we compare recent deep learning models on crop type classification on raw and preprocessed Sentinel 2 data. We concentrate on the common neural network architectures for time series, i.e., 1D-convolutions, recurrence, and the novel self-attention architecture. Our central findings are that data preprocessing still increased the overall classification performance for all models while the choice of model was less crucial. Self-attention and recurrent neural networks, by their architecture, outperformed convolutional neural networks on raw satellite time series. We explore this by a feature importance analysis based on gradient backpropagation that exploits the differentiable nature of deep learning models. Further, we qualitatively show how self-attention scores focus selectively on a few classification-relevant observations.},
  langid = {english},
  keywords = {Crop type mapping,Deep learning,Multitemporal Earth observation,Self-attention,Time series classification,Transformer,Vegetation monitoring},
  file = {C\:\\Users\\sjelic\\Zotero\\storage\\77WNTC2I\\Rußwurm and Körner - 2020 - Self-attention for raw optical Satellite Time Seri.pdf;C\:\\Users\\sjelic\\Zotero\\storage\\XL6EGUT9\\Self-attention for raw optical Satellite Time Seri.pdf}
}

@article{sadeghiSteinerTreeMethods2013,
  title = {Steiner Tree Methods for Optimal Sub-Network Identification: An Empirical Study},
  shorttitle = {Steiner Tree Methods for Optimal Sub-Network Identification},
  author = {Sadeghi, Afshin and Fröhlich, Holger},
  date = {2013-04-30},
  journaltitle = {BMC Bioinformatics},
  shortjournal = {BMC Bioinformatics},
  volume = {14},
  number = {1},
  pages = {144},
  issn = {1471-2105},
  doi = {10.1186/1471-2105-14-144},
  url = {https://doi.org/10.1186/1471-2105-14-144},
  abstract = {Analysis and interpretation of biological networks is one of the primary goals of systems biology. In this context identification of sub-networks connecting sets of seed proteins or seed genes plays a crucial role. Given that no natural node and edge weighting scheme is available retrieval of a minimum size sub-graph leads to the classical Steiner tree problem, which is known to be NP-complete. Many approximate solutions have been published and theoretically analyzed in the computer science literature, but far less is known about their practical performance in the bioinformatics field.},
  keywords = {Exact Algorithm,Minimum Span Tree,Short Path,Steiner Tree,Steiner Tree Problem},
  file = {C\:\\Users\\sjelic\\Zotero\\storage\\XBAV4DQT\\Sadeghi and Fröhlich - 2013 - Steiner tree methods for optimal sub-network ident.pdf;C\:\\Users\\sjelic\\Zotero\\storage\\U8DGYF8D\\1471-2105-14-144.html}
}

@inproceedings{saintefaregarnotSatelliteImageTime2020,
  title = {Satellite {{Image Time Series Classification With Pixel-Set Encoders}} and {{Temporal Self-Attention}}},
  booktitle = {2020 {{IEEE}}/{{CVF Conference}} on {{Computer Vision}} and {{Pattern Recognition}} ({{CVPR}})},
  author = {Sainte Fare Garnot, Vivien and Landrieu, Loic and Giordano, Sebastien and Chehata, Nesrine},
  date = {2020-06},
  pages = {12322--12331},
  issn = {2575-7075},
  doi = {10.1109/CVPR42600.2020.01234},
  abstract = {Satellite image time series, bolstered by their growing availability, are at the forefront of an extensive effort towards automated Earth monitoring by international institutions. In particular, large-scale control of agricultural parcels is an issue of major political and economic importance. In this regard, hybrid convolutional-recurrent neural architectures have shown promising results for the automated classification of satellite image time series. We propose an alternative approach in which the convolutional layers are advantageously replaced with encoders operating on unordered sets of pixels to exploit the typically coarse resolution of publicly available satellite images. We also propose to extract temporal features using a bespoke neural architecture based on self-attention instead of recurrent networks. We demonstrate experimentally that our method not only outperforms previous state-of-the-art approaches in terms of precision, but also significantly decreases processing time and memory requirements. Lastly, we release a large open-access annotated dataset as a benchmark for future work on satellite image time series.},
  eventtitle = {2020 {{IEEE}}/{{CVF Conference}} on {{Computer Vision}} and {{Pattern Recognition}} ({{CVPR}})},
  keywords = {Agriculture,Computer architecture,Feature extraction,Machine learning,Satellites,Three-dimensional displays,Time series analysis},
  file = {C\:\\Users\\sjelic\\Zotero\\storage\\A2RU4D2Q\\Sainte Fare Garnot et al. - 2020 - Satellite Image Time Series Classification With Pi.pdf;C\:\\Users\\sjelic\\Zotero\\storage\\R3ASY6JI\\9157055.html}
}

@article{salazarNoteGeneralizedSteiner2000,
  title = {A Note on the Generalized Steiner Tree Polytope},
  author = {Salazar, Juan José},
  date = {2000-03-15},
  journaltitle = {Discrete Applied Mathematics},
  shortjournal = {Discrete Applied Mathematics},
  volume = {100},
  number = {1},
  pages = {137--144},
  issn = {0166-218X},
  doi = {10.1016/S0166-218X(99)00200-0},
  url = {https://www.sciencedirect.com/science/article/pii/S0166218X99002000},
  abstract = {The generalized steiner tree problem (GSTP) is a variant of the classical Steiner tree problem (STP), in which a family of node clusters is given and the tree must span at least one node for each cluster. This note introduces a lifting procedure for obtaining polyhedral information on GSTP from polyhedral results of STP. New classes of facet-defining inequalities are presented.},
  keywords = {Facets,Polyhedron,Steiner tree},
  file = {C:\Users\sjelic\Zotero\storage\MGBBLV5M\S0166218X99002000.html}
}

@article{salhiEvolutionaryApproachCombined2019,
  title = {An Evolutionary Approach to a Combined Mixed Integer Programming Model of Seaside Operations as Arise in Container Ports},
  author = {Salhi, Abdellah and Alsoufi, Ghazwan and Yang, Xinan},
  date = {2019-01-01},
  journaltitle = {Annals of Operations Research},
  shortjournal = {Ann Oper Res},
  volume = {272},
  number = {1},
  pages = {69--98},
  issn = {1572-9338},
  doi = {10.1007/s10479-017-2539-7},
  url = {https://doi.org/10.1007/s10479-017-2539-7},
  abstract = {This paper puts forward an integrated optimisation model that combines three distinct problems, namely berth allocation, quay crane assignment, and quay crane scheduling that arise in container ports. Each one of these problems is difficult to solve in its own right. However, solving them individually leads almost surely to sub-optimal solutions. Hence, it is desirable to solve them in a combined form. The model is of the mixed-integer programming type with the objective being to minimize the tardiness of vessels and reduce the cost of berthing. Experimental results show that relatively small instances of the proposed model can be solved exactly using CPLEX. Large scale instances, however, can only be solved in reasonable times using heuristics. Here, an implementation of the genetic algorithm is considered. The effectiveness of this implementation is tested against CPLEX on small to medium size instances of the combined model. Larger size instances were also solved with the genetic algorithm, showing that this approach is capable of finding the optimal or near optimal solutions in realistic times.},
  langid = {english},
  keywords = {Berth allocation,Container terminals,Genetic algorithm,Mixed integer programming,Quay crane assignment,Quay crane scheduling},
  file = {C:\Users\sjelic\Zotero\storage\VCV6YL9P\Salhi et al. - 2019 - An evolutionary approach to a combined mixed integ.pdf}
}

@article{saulieneAutomaticPollenRecognition2019,
  title = {Automatic Pollen Recognition with the {{Rapid-E}} Particle Counter: The First-Level Procedure, Experience and next Steps},
  shorttitle = {Automatic Pollen Recognition with the {{Rapid-E}} Particle Counter},
  author = {Šaulienė, Ingrida and Šukienė, Laura and Daunys, Gintautas and Valiulis, Gediminas and Vaitkevičius, Lukas and Matavulj, Predrag and Brdar, Sanja and Panic, Marko and Sikoparija, Branko and Clot, Bernard and Crouzy, Benoît and Sofiev, Mikhail},
  date = {2019-06-28},
  journaltitle = {Atmospheric Measurement Techniques},
  volume = {12},
  number = {6},
  pages = {3435--3452},
  publisher = {Copernicus GmbH},
  issn = {1867-1381},
  doi = {10.5194/amt-12-3435-2019},
  url = {https://amt.copernicus.org/articles/12/3435/2019/},
  abstract = {{$<$}p{$><$}strong class="journal-contentHeaderColor"{$>$}Abstract.{$<$}/strong{$>$} Pollen-induced allergies are among the most prevalent non-contagious diseases, with about a quarter of the European population being sensitive to various atmospheric bioaerosols. In most European countries, pollen information is based on a weekly-cycle Hirst-type pollen trap method. This method is labour-intensive and requires narrow specialized abilities and substantial time, so that the pollen data are always delayed and subject to sampling- and counting-related uncertainties. Emerging new approaches to automatic pollen monitoring can, in principle, allow for real-time availability of the data with no human involvement.{$<$}/p{$>$} {$<$}p{$>$}The goal of the current paper is to evaluate the capabilities of the new Plair Rapid-E pollen monitor and to construct a first-level pollen recognition algorithm. The evaluation was performed for three devices located in Lithuania, Serbia and Switzerland, with independent calibration data and classification algorithms. The Rapid-E output data include multi-angle scattering images and the fluorescence spectra recorded at several times for each particle reaching the device. Both modalities of the Rapid-E output were treated with artificial neural networks (ANNs) and the results were combined to obtain the pollen type. For the first classification experiment, the monitor was challenged with a large variety of pollen types and the quality of many-to-many classification was evaluated. It was shown that in this case, both scattering- and fluorescence-based recognition algorithms fall short of acceptable quality. The combinations of these algorithms performed better, exceeding 80\&thinsp;\% accuracy for 5 out of 11 species. Fluorescence spectra showed similarities among different species, ending up with three well-resolved groups: (\mkbibemph{Alnus}, \mkbibemph{Corylus}, \mkbibemph{Betula} and \mkbibemph{Quercus}), (\mkbibemph{Salix} and \mkbibemph{Populus}) and (\mkbibemph{Festuca}, \mkbibemph{Artemisia} and \mkbibemph{Juniperus}). Within these groups, pollen is practically indistinguishable for the first-level recognition procedure. Construction of multistep algorithms with sequential discrimination of pollen inside each group seems to be one of the possible ways forward. In order to connect the classification experiment to existing technology, a short comparison with the Hirst measurements is presented and the issue of false positive pollen detections by Rapid-E is discussed.{$<$}/p{$>$}},
  langid = {english},
  file = {C:\Users\sjelic\Zotero\storage\PXM8IQLI\Šaulienė et al. - 2019 - Automatic pollen recognition with the Rapid-E part.pdf}
}

@article{saulieneAutomaticPollenRecognition2019a,
  title = {Automatic Pollen Recognition with the {{Rapid-E}} Particle Counter: The First-Level Procedure, Experience and next Steps},
  shorttitle = {Automatic Pollen Recognition with the {{Rapid-E}} Particle Counter},
  author = {Šaulienė, Ingrida and Šukienė, Laura and Daunys, Gintautas and Valiulis, Gediminas and Vaitkevičius, Lukas and Matavulj, Predrag and Brdar, Sanja and Panic, Marko and Sikoparija, Branko and Clot, Bernard and Crouzy, Benoît and Sofiev, Mikhail},
  date = {2019-06-28},
  journaltitle = {Atmospheric Measurement Techniques},
  volume = {12},
  number = {6},
  pages = {3435--3452},
  publisher = {Copernicus GmbH},
  issn = {1867-1381},
  doi = {10.5194/amt-12-3435-2019},
  url = {https://amt.copernicus.org/articles/12/3435/2019/},
  abstract = {Pollen-induced allergies are among the most prevalent non-contagious diseases, with about a quarter of the European population being sensitive to various atmospheric bioaerosols. In most European countries, pollen information is based on a weekly-cycle Hirst-type pollen trap method. This method is labour-intensive and requires narrow specialized abilities and substantial time, so that the pollen data are always delayed and subject to sampling- and counting-related uncertainties. Emerging new approaches to automatic pollen monitoring can, in principle, allow for real-time availability of the data with no human involvement.  The goal of the current paper is to evaluate the capabilities of the new Plair Rapid-E pollen monitor and to construct a first-level pollen recognition algorithm. The evaluation was performed for three devices located in Lithuania, Serbia and Switzerland, with independent calibration data and classification algorithms. The Rapid-E output data include multi-angle scattering images and the fluorescence spectra recorded at several times for each particle reaching the device. Both modalities of the Rapid-E output were treated with artificial neural networks (ANNs) and the results were combined to obtain the pollen type. For the first classification experiment, the monitor was challenged with a large variety of pollen types and the quality of many-to-many classification was evaluated. It was shown that in this case, both scattering- and fluorescence-based recognition algorithms fall short of acceptable quality. The combinations of these algorithms performed better, exceeding 80\&thinsp;\% accuracy for 5 out of 11 species. Fluorescence spectra showed similarities among different species, ending up with three well-resolved groups: (Alnus, Corylus, Betula and Quercus), (Salix and Populus) and (Festuca, Artemisia and Juniperus). Within these groups, pollen is practically indistinguishable for the first-level recognition procedure. Construction of multistep algorithms with sequential discrimination of pollen inside each group seems to be one of the possible ways forward. In order to connect the classification experiment to existing technology, a short comparison with the Hirst measurements is presented and the issue of false positive pollen detections by Rapid-E is discussed.},
  langid = {english},
  file = {C:\Users\sjelic\Zotero\storage\P5MPE4AM\Šaulienė et al. - 2019 - Automatic pollen recognition with the Rapid-E part.pdf}
}

@article{schaapLargeNeighborhoodSearch2022,
  title = {A {{Large Neighborhood Search}} for the {{Vehicle Routing Problem}} with {{Multiple Time Windows}}},
  author = {Schaap, Hendrik and Schiffer, Maximilian and Schneider, Michael and Walther, Grit},
  date = {2022-09},
  journaltitle = {Transportation Science},
  volume = {56},
  number = {5},
  pages = {1369--1392},
  publisher = {INFORMS},
  issn = {0041-1655},
  doi = {10.1287/trsc.2021.1120},
  url = {https://pubsonline.informs.org/doi/10.1287/trsc.2021.1120},
  abstract = {User-centered logistics that aim at customer satisfaction are gaining importance because of growing e-commerce and home deliveries. Customer satisfaction can be strongly increased by offering narrow delivery time windows. However, there is a tradeoff for the logistics provider because user-friendly delivery time windows might decrease operational flexibility. Against this background, we study the vehicle routing problem with multiple time windows (VRPMTW) that determines a set of optimal routes such that each customer is visited once within one out of several time windows. We present a large neighborhood search–based metaheuristic for the VRPMTW that contains a dynamic programming component to optimally select a time window for each customer on a route, and we present computationally efficient move descriptors for all search operators. We evaluate the performance of our algorithm on the Belhaiza instance set for the objectives of minimizing traveled distance and duration. For the former objective, we provide new best-known solutions for 9 of 48 instances, and for the latter, we provide new best-known solutions for 13 of 48 instances. Overall, our algorithm provides the best average solution quality over the full benchmark set among all available algorithms. Furthermore, we design new benchmark instances that reflect planning tasks in user-centered last-mile logistics. Based on these, we present managerial studies that show the benefit of our algorithm for practitioners and allow to derive insights on how to offer time windows to customers. We show that offering multiple time windows can be economically beneficial for the logistics service providers and increases customer flexibility simultaneously.},
  keywords = {dynamic programming,efficient route evaluation,multiple time windows,vehicle routing}
}

@article{schlittgenWeightedLeastsquaresApproach2011,
  title = {A Weighted Least-Squares Approach to Clusterwise Regression},
  author = {Schlittgen, Rainer},
  date = {2011-06-01},
  journaltitle = {AStA Advances in Statistical Analysis},
  shortjournal = {AStA Adv Stat Anal},
  volume = {95},
  number = {2},
  pages = {205--217},
  issn = {1863-818X},
  doi = {10.1007/s10182-011-0155-4},
  url = {https://doi.org/10.1007/s10182-011-0155-4},
  abstract = {Clusterwise regression aims to cluster data sets where the clusters are characterized by their specific regression coefficients in a linear regression model. In this paper, we propose a method for determining a partition which uses an idea of robust regression. We start with some random weighting to determine a start partition and continue in the spirit of M-estimators. The residuals for all regressions are used to assign the observations to the different groups. As target function we use the determination coefficient \$R\textasciicircum\{2\}\_\{w\}\$for the overall model. This coefficient is suitably defined for weighted regression.},
  langid = {english},
  keywords = {Bootstrap test,Clustering,Clusterwise regression,Finite mixture model,Linear regression,Robust regression,Weighted regression},
  file = {C:\Users\sjelic\Zotero\storage\T6XZHWUV\Schlittgen - 2011 - A weighted least-squares approach to clusterwise r.pdf}
}

@inproceedings{scholkopfGeneralizedRepresenterTheorem2001,
  title = {A {{Generalized Representer Theorem}}},
  booktitle = {Computational {{Learning Theory}}},
  author = {Schölkopf, Bernhard and Herbrich, Ralf and Smola, Alex J.},
  editor = {Helmbold, David and Williamson, Bob},
  date = {2001},
  pages = {416--426},
  publisher = {Springer},
  location = {Berlin, Heidelberg},
  doi = {10.1007/3-540-44581-1_27},
  abstract = {Wahba’s classical representer theorem states that the solutions of certain risk minimization problems involving an empirical risk term and a quadratic regularizer can be written as expansions in terms of the training examples. We generalize the theorem to a larger class of regularizers and empirical risk terms, and give a self-contained proof utilizing the feature space associated with a kernel. The result shows that a wide range of problems have optimal solutions that live in the finite dimensional span of the training examples mapped into feature space, thus enabling us to carry out kernel algorithms independent of the (potentially infinite) dimensionality of the feature space.},
  isbn = {978-3-540-44581-4},
  langid = {english}
}

@online{sCloudHostedEnsemble01,
  type = {chapter},
  title = {Cloud {{Hosted Ensemble Learning-Based Rental Apartment Price Prediction Model Using Stacking Technique}}},
  author = {S, Rajkumar and K, Mary Nikitha and L, Ramanathan and Ramalingam, Rajasekar and Jantwal, Mudit and S, Rajkumar and K, Mary Nikitha and L, Ramanathan and Ramalingam, Rajasekar and Jantwal, Mudit},
  date = {0001-01-01},
  publisher = {IGI Global Scientific Publishing},
  doi = {10.4018/978-1-6684-6001-6.ch015},
  url = {https://www.igi-global.com/gateway/chapter/www.igi-global.com/gateway/chapter/314147},
  abstract = {In this chapter, online rental listings of the city of Hyderabad are used as a data source for mapping house rent. Data points were scraped from one of the popular Indian rental websites www.nobroker.in. With the collected information, models of rental market dynamics were developed and evaluated us...},
  isbn = {9781668460016},
  langid = {english},
  organization = {https://services.igi-global.com/resolvedoi/resolve.aspx?doi=10.4018/978-1-6684-6001-6.ch015},
  file = {C\:\\Users\\sjelic\\Zotero\\storage\\BU9IMGE5\\S et al. - 0001 - Cloud Hosted Ensemble Learning-Based Rental Apartment Price Prediction Model Using Stacking Techniqu.pdf;C\:\\Users\\sjelic\\Zotero\\storage\\8RHJLYL3\\314147.html}
}

@online{SetConnectorProblem,
  title = {The {{Set Connector Problem}} in {{Graphs}} | {{SpringerLink}}},
  url = {https://link.springer.com/chapter/10.1007/978-3-540-72792-7_36},
  file = {C:\Users\sjelic\Zotero\storage\VPVUVKFZ\978-3-540-72792-7_36.html}
}

@book{sevauxMetaheuristics15thInternational2024,
  title = {Metaheuristics: 15th {{International Conference}}, {{MIC}} 2024, {{Lorient}}, {{France}}, {{June}} 4–7, 2024, {{Proceedings}}, {{Part I}}},
  shorttitle = {Metaheuristics},
  editor = {Sevaux, Marc and Olteanu, Alexandru-Liviu and Pardo, Eduardo G. and Sifaleras, Angelo and Makboul, Salma},
  date = {2024},
  series = {Lecture {{Notes}} in {{Computer Science}}},
  volume = {14753},
  publisher = {Springer Nature Switzerland},
  location = {Cham},
  doi = {10.1007/978-3-031-62912-9},
  url = {https://link.springer.com/10.1007/978-3-031-62912-9},
  isbn = {978-3-031-62911-2 978-3-031-62912-9},
  langid = {english},
  keywords = {artificial intelligence,combinatorial optimization,computer science,evolutionary algorithms,genetic algorithms,graph theory,optimization,optimization problems,simulated annealing},
  file = {C:\Users\sjelic\Zotero\storage\NBGL54UG\Sevaux et al. - 2024 - Metaheuristics 15th International Conference, MIC.pdf}
}

@article{sinhaEffectiveMovieRecommender2021a,
  title = {An Effective Movie Recommender System Enhanced with Time Series Analysis of User Rating Behaviour},
  author = {Sinha, Bam Bahadur and Dhanalakshmi, R.},
  date = {2021-07-28},
  journaltitle = {International Journal of Mathematics in Operational Research},
  publisher = {Inderscience Publishers (IEL)},
  url = {https://www.inderscienceonline.com/doi/10.1504/IJMOR.2021.116966},
  abstract = {Recommender system aims at improvising user satisfaction by taking decision on what movie or item to recommend next. Over time though, learners and learning behaviours shift regularly. This paper introduces a novel behaviour-inspired suggestion algorithm named the TimeFly-PPSE algorithm, which operates on the concept of changing user's motives around time. The suggested model takes temporal knowledge into account and monitors the progression of consumers and items that are useful in providing adequate recommendations. The latter outlines a framework that enrolls the user's shifting behaviour to include guidance for personalisation. TimeFly's findings are contrasted with those of other well-known algorithms. Simulation test on 100K MovieLens dataset shows that utilising TimeFly contributes to recommendations that are exceptionally efficient and reliable.},
  langid = {english},
  file = {C:\Users\sjelic\Zotero\storage\DV68WDS7\IJMOR.2021.html}
}

@thesis{slavikApproximationAlgorithmsSet1998,
  type = {phdthesis},
  title = {Approximation Algorithms for Set Cover and Related Problems},
  author = {Slavik, Petr},
  date = {1998},
  institution = {State University of New York at Buffalo},
  location = {USA},
  abstract = {In this thesis, we analyze several known and newly designed algorithms for approximating optimal solutions to NP-hard optimization problems. We give a new analysis of the greedy algorithm for approximating the S scET C scOVER and P scARTIAL S scET scCOVER problems obtaining significantly improved performance bounds. We also give a first approximation algorithm with a non-trivial performance bound for the E scRRAND S scCHEDULING and T scREE C scOVER problems, known also as the G scENERALIZED T scRAVELING S scALESMAN and G scROUP S scTEINER T scREE problems.The main results of this thesis first appeared in my papers (87), (89), (91), and (90); and in my technical reports (86) and (88).},
  pagetotal = {158},
  annotation = {AAI9833643\\
ISBN-10: 059186777X}
}

@online{SolvingSteinerTree,
  title = {Solving {{Steiner}} Tree Problems in Graphs to Optimality - {{Koch}} - 1998 - {{Networks}} - {{Wiley Online Library}}},
  url = {https://onlinelibrary.wiley.com/doi/abs/10.1002/%28SICI%291097-0037%28199810%2932%3A3%3C207%3A%3AAID-NET5%3E3.0.CO%3B2-O},
  file = {C:\Users\sjelic\Zotero\storage\VWV6RTH7\(SICI)1097-0037(199810)323207AID-NET53.0.html}
}

@article{spathAlgorithm39Clusterwise1979,
  title = {Algorithm 39 {{Clusterwise}} Linear Regression},
  author = {Späth, H.},
  date = {1979-12-01},
  journaltitle = {Computing},
  shortjournal = {Computing},
  volume = {22},
  number = {4},
  pages = {367--373},
  issn = {1436-5057},
  doi = {10.1007/BF02265317},
  url = {https://doi.org/10.1007/BF02265317},
  abstract = {The combinatorial problem of clusterwise discrete linear approximation is defined as finding a given number of clusters of observations such that the overall sum of error sum of squares within those clusters becomes a minimum. The FORTRAN implementation of a heuristic solution method and a numerical example are given.},
  langid = {english},
  keywords = {Combinatorial Problem,Computational Mathematic,Linear Approximation,Linear Regression,Solution Method},
  file = {C:\Users\sjelic\Zotero\storage\QUGJRGCE\Spiith - Algorithm 39 Clusterwise linear regression.pdf}
}

@article{spathFastAlgorithmClusterwise1982,
  title = {A Fast Algorithm for Clusterwise Linear Regression},
  author = {Späth, H.},
  date = {1982-06-01},
  journaltitle = {Computing},
  shortjournal = {Computing},
  volume = {29},
  number = {2},
  pages = {175--181},
  issn = {1436-5057},
  doi = {10.1007/BF02249940},
  url = {https://doi.org/10.1007/BF02249940},
  abstract = {A fast implementation of a formerly [5] published algorithm is given.},
  langid = {english},
  keywords = {62H30,62J05,65D10,65F20,Artificial Intelligence,Cluster analysis,linear regression}
}

@article{splietDiscreteTimeWindow2015,
  title = {The Discrete Time Window Assignment Vehicle Routing Problem},
  author = {Spliet, Remy and Desaulniers, Guy},
  date = {2015-07-16},
  journaltitle = {European Journal of Operational Research},
  shortjournal = {European Journal of Operational Research},
  volume = {244},
  number = {2},
  pages = {379--391},
  issn = {0377-2217},
  doi = {10.1016/j.ejor.2015.01.020},
  url = {https://www.sciencedirect.com/science/article/pii/S0377221715000405},
  abstract = {In this paper we introduce the discrete time window assignment vehicle routing problem (DTWAVRP) that can be viewed as a two-stage stochastic optimization problem. Given a set of customers that must be visited on the same day regularly within some period of time, the first-stage decisions are to assign to each customer a time window from a set of candidate time windows before demand is known. In the second stage, when demand is revealed for each day of the time period, vehicle routes satisfying vehicle capacity and the assigned time windows are constructed. The objective of the DTWAVRP is to minimize the expected total transportation cost. To solve this problem, we develop an exact branch-price-and-cut algorithm and derive from it five column generation heuristics that allow to solve larger instances than those solved by the exact algorithm. We illustrate the performance of these algorithms by means of computational experiments performed on randomly generated instances.},
  keywords = {Column generation,Time window assignment,Uncertain demand,Vehicle routing},
  file = {C:\Users\sjelic\Zotero\storage\ZTAURE5H\S0377221715000405.html}
}

@article{splietTimeWindowAssignment2015,
  title = {The {{Time Window Assignment Vehicle Routing Problem}}},
  author = {Spliet, Remy and Gabor, Adriana F.},
  date = {2015-11},
  journaltitle = {Transportation Science},
  volume = {49},
  number = {4},
  pages = {721--731},
  publisher = {INFORMS},
  issn = {0041-1655},
  doi = {10.1287/trsc.2013.0510},
  url = {https://pubsonline.informs.org/doi/10.1287/trsc.2013.0510},
  abstract = {In this paper we introduce the time window assignment vehicle routing problem (TWAVRP). In this problem, time windows have to be assigned before demand is known. Next, a realization of demand is revealed, and a vehicle routing schedule is made that satisfies the assigned time windows. The objective is to minimize the expected traveling costs. We propose a branch-price-and-cut algorithm to solve the TWAVRP to optimality. We provide results of computational experiments performed using this algorithm. Finally, we offer insight on the value of an exact approach for the TWAVRP by comparing the optimal solution to the solution found by assigning time windows based on solving a vehicle routing problem with time windows with average demand.},
  keywords = {pricing problem with linear node costs,time window assignment,vehicle routing problem},
  file = {C\:\\Users\\sjelic\\Zotero\\storage\\NHUL2595\\trsc2E20132E0510.pdf;C\:\\Users\\sjelic\\Zotero\\storage\\ZBP4E3W7\\Spliet and Gabor - 2015 - The Time Window Assignment Vehicle Routing Problem.pdf}
}

@article{splietTimeWindowAssignment2017,
  title = {The {{Time Window Assignment Vehicle Routing Problem}} with {{Time-Dependent Travel Times}}},
  author = {Spliet, Remy and Dabia, Said and Van Woensel, Tom},
  date = {2017-03-21},
  journaltitle = {Transportation Science},
  shortjournal = {Transportation Science},
  volume = {52},
  doi = {10.1287/trsc.2016.0705},
  abstract = {In this paper, we introduce the time window assignment vehicle routing problem (TWAVRP) with time-dependent travel times. It is the problem of assigning time windows to customers before their demand is known and creating vehicle routes adhering to these time windows after demand becomes known. The goal is to assign the time windows in such a way that the expected transportation costs are minimized.We develop a branch-price-And-cut algorithm to solve this problem to optimality. The pricing problem that has to be solved is a newvariant of the shortest path problem, which includes a capacity constraint, time-dependent travel times, time window constraints on both the nodes and on the arcs, and linear node costs. For solving the pricing problem, we develop an exact labeling algorithm and a tabu search heuristic. Furthermore, we present new valid inequalities, which are specifically designed for the TWAVRP with time-dependent travel times. Finally, we present results of numerical experiments to illustrate the performance of the algorithm.},
  file = {C:\Users\sjelic\Zotero\storage\2ZAZHT8D\Spliet et al. - 2017 - The Time Window Assignment Vehicle Routing Problem with Time-Dependent Travel Times.pdf}
}

@article{stakicReducedVariableNeighborhood2021,
  title = {A {{Reduced Variable Neighborhood Search Approach}} to the {{Heterogeneous Vector Bin Packing Problem}}},
  author = {Stakić, Đorđe and Živković, Miodrag and Anokić, Ana},
  date = {2021-12-16},
  journaltitle = {Information Technology and Control},
  volume = {50},
  number = {4},
  pages = {808--826},
  issn = {2335-884X},
  doi = {10.5755/j01.itc.50.4.29009},
  url = {https://itc.ktu.lt/index.php/ITC/article/view/29009},
  abstract = {The two-dimensional heterogeneous vector bin packing problem (2DHet-VBPP) consists of packing the set of items into the set of various type bins, respecting their two resource limits. The problem is to minimize the total cost of all bins. The problem, known to be NP-hard, can be formulated as a pure integer linear program, but optimal solutions can be obtained by the CPLEX Optimizer engine only for small instances. This paper proposes a metaheuristic approach to the 2DHet-VBPP, based on Reduced variable neighborhood search (RVNS). All RVNS elements are adapted to the considered problem and many procedures are designed to improve efficiency of the method. As the Two-dimensional Homogeneous-VBPP (2DHom-VBPP) is more often treated, we considered also a special version of the RVNS algorithm to solve the 2DHom-VBPP. The results obtained and compared to both CPLEX results and results on benchmark instances from literature, justify the use of the RVNS algorithm to solve large instances of these optimization problems.},
  issue = {4},
  langid = {english},
  keywords = {Metaheuristics},
  file = {C:\Users\sjelic\Zotero\storage\RU5HBP2M\Stakić et al. - 2021 - A Reduced Variable Neighborhood Search Approach to.pdf}
}

@article{steurerMetricsEvaluatingPerformance2021,
  title = {Metrics for Evaluating the Performance of Machine Learning Based Automated Valuation Models},
  author = {Steurer, Miriam and Hill, Robert J. and Pfeifer, Norbert},
  date = {2021-04-03},
  journaltitle = {Journal of Property Research},
  volume = {38},
  number = {2},
  pages = {99--129},
  publisher = {Routledge},
  issn = {0959-9916},
  doi = {10.1080/09599916.2020.1858937},
  url = {https://doi.org/10.1080/09599916.2020.1858937},
  abstract = {Automated Valuation Models (AVMs) based on Machine Learning (ML) algorithms are widely used for predicting house prices. While there is consensus in the literature that cross-validation (CV) should be used for model selection in this context, the interdisciplinary nature of the subject has made it hard to reach consensus over which metrics to use at each stage of the CV exercise. We collect 48 metrics (from the AVM literature and elsewhere) and classify them into seven groups according to their structure. Each of these groups focuses on a particular aspect of the error distribution. Depending on the type of data and the purpose of the AVM, the needs of users may be met by some classes, but not by others. In addition, we show in an empirical application how the choice of metric can influence the choice of model, by applying each metric to evaluate five commonly used AVM models. Finally – since it is not always practicable to produce 48 different performance metrics – we provide a short list of 7 metrics that are well suited to evaluate AVMs. These metrics satisfy a symmetry condition that we find is important for AVM performance, and can provide a good overall model performance ranking.},
  keywords = {automated valuation,house price prediction,machine learning,model selection,Performance metrics},
  file = {C:\Users\sjelic\Zotero\storage\4Q2LWVAN\Steurer et al. - 2021 - Metrics for evaluating the performance of machine .pdf}
}

@article{sunFindingGroupSteiner2021,
  title = {Finding Group {{Steiner}} Trees in Graphs with Both Vertex and Edge Weights},
  author = {Sun, Yahui and Xiao, Xiaokui and Cui, Bin and Halgamuge, Saman and Lappas, Theodoros and Luo, Jun},
  date = {2021-03},
  journaltitle = {Proceedings of the VLDB Endowment},
  shortjournal = {Proc. VLDB Endow.},
  volume = {14},
  number = {7},
  pages = {1137--1149},
  issn = {2150-8097},
  doi = {10.14778/3450980.3450982},
  url = {https://dl.acm.org/doi/10.14778/3450980.3450982},
  abstract = {Given an undirected graph and a number of vertex groups, the group Steiner tree problem is to find a tree such that (i) this tree contains at least one vertex in each vertex group; and (ii) the sum of vertex and edge weights in this tree is minimized. Solving this problem is useful in various scenarios, ranging from social networks to knowledge graphs. Most existing work focuses on solving this problem in vertex-unweighted graphs, and not enough work has been done to solve this problem in graphs with both vertex and edge weights. Here, we develop several algorithms to address this issue. Initially, we extend two algorithms from vertex-unweighted graphs to vertex- and edge-weighted graphs. The first one has no approximation guarantee, but often produces good solutions in practice. The second one has an approximation guarantee of |Γ| − 1, where |Γ| is the number of vertex groups. Since the extended (|Γ| − 1)approximation algorithm is too slow when all vertex groups are large, we develop two new (|Γ| − 1)-approximation algorithms that overcome this weakness. Furthermore, by employing a dynamic programming approach, we develop another (|Γ| −ℎ +1)-approximation algorithm, where ℎ is a parameter between 2 and |Γ|. Experiments show that, while no algorithm is the best in all cases, our algorithms considerably outperform the state of the art in many scenarios.},
  langid = {english},
  file = {C:\Users\sjelic\Zotero\storage\L8B3SHSJ\Sun et al. - 2021 - Finding group Steiner trees in graphs with both ve.pdf}
}

@article{sunNodeweightedSteinerTree2017,
  title = {The Node-Weighted {{Steiner}} Tree Approach to Identify Elements of Cancer-Related Signaling Pathways},
  author = {Sun, Yahui and Ma, Chenkai and Halgamuge, Saman},
  date = {2017-12-28},
  journaltitle = {BMC Bioinformatics},
  shortjournal = {BMC Bioinformatics},
  volume = {18},
  number = {16},
  pages = {551},
  issn = {1471-2105},
  doi = {10.1186/s12859-017-1958-4},
  url = {https://doi.org/10.1186/s12859-017-1958-4},
  abstract = {Cancer constitutes a momentous health burden in our society. Critical information on cancer may be hidden in its signaling pathways. However, even though a large amount of money has been spent on cancer research, some critical information on cancer-related signaling pathways still remains elusive. Hence, new works towards a complete understanding of cancer-related signaling pathways will greatly benefit the prevention, diagnosis, and treatment of cancer.},
  keywords = {Big data,Bioinformatics,Data mining,Systems biology},
  file = {C\:\\Users\\sjelic\\Zotero\\storage\\VLDHTNN8\\Sun et al. - 2017 - The node-weighted Steiner tree approach to identif.pdf;C\:\\Users\\sjelic\\Zotero\\storage\\RLBJ533I\\s12859-017-1958-4.html}
}

@article{sunNondestructiveDetectionForeign2022,
  title = {Non-Destructive Detection for Foreign Bodies of Tea Stalks in Finished Tea Products Using Terahertz Spectroscopy and Imaging},
  author = {Sun, Xudong and Cui, Dongdong and Shen, Yun and Li, Wenping and Wang, Jiahua},
  date = {2022-03-01},
  journaltitle = {Infrared Physics \& Technology},
  shortjournal = {Infrared Physics \& Technology},
  volume = {121},
  pages = {104018},
  issn = {1350-4495},
  doi = {10.1016/j.infrared.2021.104018},
  url = {https://www.sciencedirect.com/science/article/pii/S135044952100390X},
  abstract = {Foreign bodies e.g. tea stalks are easily to mix into tea leaves during picking up tea leaves with machine harvester. Tea stems have become a major physical pollutant for finishing tea products, and they cannot be effectively detected with machine vision in the visible region. In the current study, terahertz time domain spectroscopy (THz-TDS) and imaging was employed to rapidly and non-destructively detect tea stalk foreign bodies in tea leaves. With the input variables of THz time-domain signal and frequency-domain absorption coefficients, the K nearest neighbor (KNN) qualitative discriminant models were developed with combination of baseline correction algorithms of adaptive iteratively reweighted penalized least squares (AirPLS), asymmetric least squares (AsLS), background correction (Backcor) and baseline estimation and denoising with sparsity (BEADS). Results showed that the AirPLS-KNN model with the input vector of THz time-domain signals presented the best performance, with the accuracy rate of prediction and recall rate reaching 97.3\% and 0.96, respectively. With 0.2~mm step length of both the~X~and Y directions, the THz transmission imaging was scanned to obtain the THz image of 150X130 pixel2 whose resolution and imaging duration were 1.09~mm and 4.4~min respectively, so the outline of tea stalks could be well identified. In conclusion, the THz-TDS and imaging technology serves as a new means for non-destructive detection of tea stalk foreign bodies in tea leaves.},
  keywords = {Foreign body,Imaging,Spectroscopy,Tea,Terahertz},
  file = {C\:\\Users\\sljel_0obp3t6\\OneDrive\\Radna površina\\1-s2.0-S135044952100390X-main.pdf;C\:\\Users\\sjelic\\Zotero\\storage\\YVI6AWKV\\S135044952100390X.html}
}

@article{sunPhysaruminspiredPrizecollectingSteiner2016,
  title = {A Physarum-Inspired Prize-Collecting Steiner Tree Approach to Identify Subnetworks for Drug Repositioning},
  author = {Sun, Yahui and Hameed, Pathima Nusrath and Verspoor, Karin and Halgamuge, Saman},
  date = {2016-12-05},
  journaltitle = {BMC Systems Biology},
  shortjournal = {BMC Systems Biology},
  volume = {10},
  number = {5},
  pages = {128},
  issn = {1752-0509},
  doi = {10.1186/s12918-016-0371-3},
  url = {https://doi.org/10.1186/s12918-016-0371-3},
  abstract = {Drug repositioning can reduce the time, costs and risks of drug development by identifying new therapeutic effects for known drugs. It is challenging to reposition drugs as pharmacological data is large and complex. Subnetwork identification has already been used to simplify the visualization and interpretation of biological data, but it has not been applied to drug repositioning so far. In this paper, we fill this gap by proposing a new Physarum-inspired Prize-Collecting Steiner Tree algorithm to identify subnetworks for drug repositioning.},
  keywords = {Big data,Drug similarity network,Physarum polycephalum,Steiner tree problem,Subnetwork identification},
  file = {C:\Users\sjelic\Zotero\storage\YIYQVUUB\Sun et al. - 2016 - A physarum-inspired prize-collecting steiner tree .pdf}
}

@inproceedings{sutskeverSequenceSequenceLearning2014,
  title = {Sequence to {{Sequence Learning}} with {{Neural Networks}}},
  booktitle = {Advances in {{Neural Information Processing Systems}}},
  author = {Sutskever, Ilya and Vinyals, Oriol and Le, Quoc V},
  date = {2014},
  volume = {27},
  publisher = {Curran Associates, Inc.},
  url = {https://proceedings.neurips.cc/paper/2014/hash/a14ac55a4f27472c5d894ec1c3c743d2-Abstract.html},
  abstract = {Deep Neural Networks (DNNs) are powerful models that have achieved excellent performance on difficult learning tasks. Although DNNs work well whenever large labeled training sets are available, they cannot be used to map sequences to sequences. In this paper, we present a general end-to-end approach to sequence learning that makes minimal assumptions on the sequence structure. Our method uses a multilayered Long Short-Term Memory (LSTM) to map the input sequence to a vector of a fixed dimensionality, and then another deep LSTM to decode the target sequence from the vector. Our main result is that on an English to French translation task from the WMT-14 dataset, the translations produced by the LSTM achieve a BLEU score of 34.8 on the entire test set, where the LSTM's BLEU score was penalized on out-of-vocabulary words. Additionally, the LSTM did not have difficulty on long sentences. For comparison, a phrase-based SMT system achieves a BLEU score of 33.3 on the same dataset. When we used the LSTM to rerank the 1000 hypotheses produced by the aforementioned SMT system, its BLEU score increases to 36.5, which is close to the previous state of the art. The LSTM also learned sensible phrase and sentence representations that are sensitive to word order and are relatively invariant to the active and the passive voice. Finally, we found that reversing the order of the words in all source sentences (but not target sentences) improved the LSTM's performance markedly, because doing so introduced many short term dependencies between the source and the target sentence which made the optimization problem easier.},
  file = {C:\Users\sjelic\Zotero\storage\GNZDDH4S\Sutskever et al. - 2014 - Sequence to Sequence Learning with Neural Networks.pdf}
}

@article{tanVehicleRoutingProblem2021,
  title = {The {{Vehicle Routing Problem}}: {{State-of-the-Art Classification}} and {{Review}}},
  shorttitle = {The {{Vehicle Routing Problem}}},
  author = {Tan, Shi-Yi and Yeh, Wei-Chang},
  date = {2021-01},
  journaltitle = {Applied Sciences},
  volume = {11},
  number = {21},
  pages = {10295},
  publisher = {Multidisciplinary Digital Publishing Institute},
  issn = {2076-3417},
  doi = {10.3390/app112110295},
  url = {https://www.mdpi.com/2076-3417/11/21/10295},
  abstract = {Transportation planning has been established as a key topic in the literature and social production practices. An increasing number of researchers are studying vehicle routing problems (VRPs) and their variants considering real-life applications and scenarios. Furthermore, with the rapid growth in the processing speed and memory capacity of computers, various algorithms can be used to solve increasingly complex instances of VRPs. In this study, we analyzed recent literature published between 2019 and August of 2021 using a taxonomic framework. We reviewed recent research according to models and solutions, and divided models into three categories of customer-related, vehicle-related, and depot-related models. We classified solution algorithms into exact, heuristic, and meta-heuristic algorithms. The main contribution of our study is a classification table that is available online as Appendix A. This classification table should enable future researchers to find relevant literature easily and provide readers with recent trends and solution methodologies in the field of VRPs and some well-known variants.},
  issue = {21},
  langid = {english},
  keywords = {exact methods,heuristics,literature review,meta-heuristics,taxonomy,vehicle routing problem},
  file = {C:\Users\sjelic\Zotero\storage\HKC7Q35E\Tan и Yeh - 2021 - The Vehicle Routing Problem State-of-the-Art Classification and Review.pdf}
}

@article{tavassoliComparisonKrigingArtificial2022,
  title = {Comparison of {{Kriging}} and Artificial Neural Network Models for the Prediction of Spatial Data},
  author = {Tavassoli, Abbas and Waghei, Yadollah and Nazemi, Alireza},
  date = {2022-01-22},
  journaltitle = {Journal of Statistical Computation and Simulation},
  volume = {92},
  number = {2},
  pages = {352--369},
  publisher = {Taylor \& Francis},
  issn = {0094-9655},
  doi = {10.1080/00949655.2021.1961140},
  url = {https://doi.org/10.1080/00949655.2021.1961140},
  abstract = {The prediction of a spatial variable is of particular importance when analyzing spatial data. The main objective of this study is to evaluate and compare the performance of several prediction-based methods in spatial prediction through a simulation study. The studied methods include ordinary Kriging (OK), along with several neural network methods including Multi-Layer Perceptron network (MLP), Ensemble Neural Networks (ENN), and Radial Basis Function (RBF) network. We simulated several spatial datasets with three different scenarios due to changes in data stationarity and isotropy. The performance of methods was evaluated using the Root Mean Square Error (RMSE), Mean Absolute Error (MAE) and Concordance Correlation Coefficient (CCC) indexes. Although the results of the simulation study revealed that the performance of the neural network in spatial prediction is weaker than the Kriging method, but it can still be a good competitor for Kriging.},
  keywords = {Artificial neural network,Kriging,multilayer perceptron,radial basis function,simulation,spatial prediction}
}

@article{taylorPostselectionInferenceL1penalized2018,
  title = {Post-Selection Inference for ℓ1-Penalized Likelihood Models},
  author = {Taylor, Jonathan and Tibshirani, Robert},
  date = {2018-03},
  journaltitle = {Canadian Journal of Statistics},
  shortjournal = {Can. J. Statistics},
  volume = {46},
  number = {1},
  pages = {41--61},
  issn = {03195724},
  doi = {10.1002/cjs.11313},
  url = {https://onlinelibrary.wiley.com/doi/10.1002/cjs.11313},
  abstract = {We present a new method for post-selection inference for ℓ1 (lasso)-penalized likelihood models, including generalized regression models. Our approach generalizes the post-selection framework presented in Lee et al. (2013). The method provides p-values and confidence intervals that are asymptotically valid, conditional on the inherent selection done by the lasso. We present applications of this work to (regularized) logistic regression, Cox's proportional hazards model and the graphical lasso. We do not provide rigorous proofs here of the claimed results, but rather conceptual and theoretical sketches.},
  langid = {english},
  file = {C:\Users\sjelic\Zotero\storage\BFRBMITG\Taylor and Tibshirani - 2018 - Post-selection inference for ℓ1-penalized likeliho.pdf}
}

@article{tibshiraniRegressionShrinkageSelection1996,
  title = {Regression {{Shrinkage}} and {{Selection Via}} the {{Lasso}}},
  author = {Tibshirani, Robert},
  date = {1996},
  journaltitle = {Journal of the Royal Statistical Society: Series B (Methodological)},
  volume = {58},
  number = {1},
  pages = {267--288},
  issn = {2517-6161},
  doi = {10.1111/j.2517-6161.1996.tb02080.x},
  url = {https://onlinelibrary.wiley.com/doi/abs/10.1111/j.2517-6161.1996.tb02080.x},
  abstract = {We propose a new method for estimation in linear models. The ‘lasso’ minimizes the residual sum of squares subject to the sum of the absolute value of the coefficients being less than a constant. Because of the nature of this constraint it tends to produce some coefficients that are exactly 0 and hence gives interpretable models. Our simulation studies suggest that the lasso enjoys some of the favourable properties of both subset selection and ridge regression. It produces interpretable models like subset selection and exhibits the stability of ridge regression. There is also an interesting relationship with recent work in adaptive function estimation by Donoho and Johnstone. The lasso idea is quite general and can be applied in a variety of statistical models: extensions to generalized regression models and tree-based models are briefly described.},
  langid = {english},
  keywords = {quadratic programming,regression,shrinkage,subset selection},
  file = {C\:\\Users\\sjelic\\Zotero\\storage\\CNGQ6H6A\\Tibshirani - 1996 - Regression Shrinkage and Selection Via the Lasso.pdf;C\:\\Users\\sjelic\\Zotero\\storage\\G8WE4CYE\\j.2517-6161.1996.tb02080.html}
}

@article{turkogullariOptimalBerthAllocation2016,
  title = {Optimal Berth Allocation, Time-Variant Quay Crane Assignment and Scheduling with Crane Setups in Container Terminals},
  author = {Türkoğulları, Yavuz B. and Taşkın, Z. Caner and Aras, Necati and Altınel, İ. Kuban},
  date = {2016-11-01},
  journaltitle = {European Journal of Operational Research},
  shortjournal = {European Journal of Operational Research},
  volume = {254},
  number = {3},
  pages = {985--1001},
  issn = {0377-2217},
  doi = {10.1016/j.ejor.2016.04.022},
  url = {https://www.sciencedirect.com/science/article/pii/S0377221716302417},
  abstract = {There has been a dramatic increase in world’s container traffic during the last thirty years. As a consequence, the efficient management of container terminals has become a crucial issue. In this work we concentrate on the integrated seaside operations, namely the integration of berth allocation, quay crane assignment and quay crane scheduling problems. First, we formulate a mixed-integer linear program whose exact solution gives optimal berthing positions and berthing times of the vessels, along with their crane schedules during their stay at the quay. Then, we propose an efficient cutting plane algorithm based on a decomposition scheme. Our approach deals with berthing positions of the vessels and their assigned number of cranes in each time period in a master problem, and seeks the corresponding optimal crane schedule by solving a subproblem. We prove that the crane scheduling subproblem is NP-complete under general cost settings, but can be solved in polynomial time for certain special cases. Our computational study shows that our new formulation and proposed solution method yield optimal solutions for realistic-sized instances.},
  keywords = {Berth allocation,Container terminal management,Crane assignment,Crane scheduling,Integer programming},
  file = {C\:\\Users\\sjelic\\Zotero\\storage\\WKHHTI82\\Türkoğulları et al. - 2016 - Optimal berth allocation, time-variant quay crane .pdf;C\:\\Users\\sjelic\\Zotero\\storage\\87M52DEW\\S0377221716302417.html}
}

@inproceedings{vaswaniAttentionAllYou2017,
  title = {Attention Is All You Need},
  booktitle = {Proceedings of the 31st {{International Conference}} on {{Neural Information Processing Systems}}},
  author = {Vaswani, Ashish and Shazeer, Noam and Parmar, Niki and Uszkoreit, Jakob and Jones, Llion and Gomez, Aidan N. and Kaiser, Łukasz and Polosukhin, Illia},
  date = {2017-12-04},
  series = {{{NIPS}}'17},
  pages = {6000--6010},
  publisher = {Curran Associates Inc.},
  location = {Red Hook, NY, USA},
  abstract = {The dominant sequence transduction models are based on complex recurrent or convolutional neural networks that include an encoder and a decoder. The best performing models also connect the encoder and decoder through an attention mechanism. We propose a new simple network architecture, the Transformer, based solely on attention mechanisms, dispensing with recurrence and convolutions entirely. Experiments on two machine translation tasks show these models to be superior in quality while being more parallelizable and requiring significantly less time to train. Our model achieves 28.4 BLEU on the WMT 2014 English-to-German translation task, improving over the existing best results, including ensembles, by over 2 BLEU. On the WMT 2014 English-to-French translation task, our model establishes a new single-model state-of-the-art BLEU score of 41.0 after training for 3.5 days on eight GPUs, a small fraction of the training costs of the best models from the literature.},
  isbn = {978-1-5108-6096-4},
  file = {C:\Users\sjelic\Zotero\storage\48MZCQ67\Vaswani et al. - 2017 - Attention is all you need.pdf}
}

@book{venkateswaraDomainAdaptationComputer2020a,
  title = {Domain {{Adaptation}} in {{Computer Vision}} with {{Deep Learning}}},
  editor = {Venkateswara, Hemanth and Panchanathan, Sethuraman},
  date = {2020},
  publisher = {Springer International Publishing},
  location = {Cham},
  doi = {10.1007/978-3-030-45529-3},
  url = {http://link.springer.com/10.1007/978-3-030-45529-3},
  isbn = {978-3-030-45528-6 978-3-030-45529-3},
  langid = {english},
  file = {C:\Users\sjelic\Zotero\storage\SXIBZH7H\Venkateswara and Panchanathan - 2020 - Domain Adaptation in Computer Vision with Deep Lea.pdf}
}

@incollection{venkateswaraIntroductionDomainAdaptation2020,
  title = {Introduction to {{Domain Adaptation}}},
  booktitle = {Domain {{Adaptation}} in {{Computer Vision}} with {{Deep Learning}}},
  author = {Venkateswara, Hemanth and Panchanathan, Sethuraman},
  editor = {Venkateswara, Hemanth and Panchanathan, Sethuraman},
  date = {2020},
  pages = {3--21},
  publisher = {Springer International Publishing},
  location = {Cham},
  doi = {10.1007/978-3-030-45529-3_1},
  url = {https://doi.org/10.1007/978-3-030-45529-3_1},
  abstract = {This chapter provides a formal introduction to transfer learning. We define transfer learning and provide examples of different forms of transfer learning in machine learning including domain adaptation. We outline different forms of domain adaptation and derive it's performance bounds. The final section presents a brief description of the chapters in the book.},
  isbn = {978-3-030-45529-3}
}

@article{wangExploringPotentialMultisource2022,
  title = {Exploring the Potential of Multi-Source Unsupervised Domain Adaptation in Crop Mapping Using {{Sentinel-2}} Images},
  author = {Wang, Yumiao and Feng, Luwei and Sun, Weiwei and Zhang, Zhou and Zhang, Hanyu and Yang, Gang and Meng, Xiangchao},
  date = {2022-12-31},
  journaltitle = {GIScience \& Remote Sensing},
  volume = {59},
  number = {1},
  pages = {2247--2265},
  publisher = {Taylor \& Francis},
  issn = {1548-1603},
  doi = {10.1080/15481603.2022.2156123},
  url = {https://doi.org/10.1080/15481603.2022.2156123},
  abstract = {Accurate crop mapping is critical for agricultural applications. Although studies have combined deep learning methods and time-series satellite images to crop classification with satisfactory results, most of them focused on supervised methods, which are usually applicable to a specific domain and lose their validity in new domains. Unsupervised domain adaptation (UDA) was proposed to solve this limitation by transferring knowledge from source domains with labeled samples to target domains with unlabeled samples. Particularly, multi-source UDA (MUDA) is a powerful extension that leverages knowledge from multiple source domains and can achieve better results in the target domain than single-source UDA (SUDA). However, few studies have explored the potential of MUDA for crop mapping. This study proposed a MUDA crop classification model (MUCCM) for unsupervised crop mapping. Specifically, 11 states in the U.S. were selected as the multi-source domains, and three provinces in Northeast China were selected as individual target domains. Ten spectral bands and five vegetation indexes were collected at a 10-day interval from time-series Sentinel-2 images to build the MUCCM. Subsequently, a SUDA model Domain Adversarial Neural Network (DANN) and two direct transfer methods, namely, the deep neural network and random forest, were constructed and compared with the MUCCM. The results indicated that the UDA models outperformed the direct transfer models significantly, and the MUCCM was superior to the DANN, achieving the highest classification accuracy (OA{$>$}85\%) in each target domain. In addition, the MUCCM also performed best in in-season forecasting and crop mapping. This study is the first to apply a MUDA to crop classification and demonstrate a novel, effective solution for high-performance crop mapping in regions without labeled samples.},
  keywords = {Crop mapping,deep learning,multi-source unsupervised domain adaptation,time-series remote sensing,transfer learning},
  file = {C:\Users\sjelic\Zotero\storage\6R2VJDU9\Wang и сар. - 2022 - Exploring the potential of multi-source unsupervis.pdf}
}

@inproceedings{wangKernelPathKernelized2007,
  title = {The {{Kernel Path}} in {{Kernelized LASSO}}},
  author = {Wang, G. and Yeung, D. and Lochovsky, F.},
  date = {2007-03-11},
  url = {https://www.semanticscholar.org/paper/The-Kernel-Path-in-Kernelized-LASSO-Wang-Yeung/79f4dbb9f8de932567948d4b11ac1b5bcd9d13c2},
  abstract = {Kernel methods implicitly map data points from the input space to some feature space where even relatively simple algorithms such as linear methods can deliver very impressive performance. Of crucial importance though is the choice of the kernel function, which determines the mapping between the input space and the feature space. The past few years have seen many efforts in learning either the kernel function or the kernel matrix. In this paper, we study the problem of learning the kernel hyperparameter in the context of the kernelized LASSO regression model. Specifically, we propose a solution path algorithm with respect to the hyperparameter of the kernel function. As the kernel hyperparameter changes its value, the solution path can be traced exactly without having to train the model multiple times. As a result, the optimal solution can be identified efficiently. Some simulation results will be presented to demonstrate the effectiveness of our proposed kernel path algorithm.},
  eventtitle = {International {{Conference}} on {{Artificial Intelligence}} and {{Statistics}}},
  file = {C:\Users\sjelic\Zotero\storage\DIBDV49I\Wang et al. - 2007 - The Kernel Path in Kernelized LASSO.pdf}
}

@inproceedings{wilsonMultiSourceDeepDomain2020,
  title = {Multi-{{Source Deep Domain Adaptation}} with {{Weak Supervision}} for {{Time-Series Sensor Data}}},
  booktitle = {Proceedings of the 26th {{ACM SIGKDD International Conference}} on {{Knowledge Discovery}} \& {{Data Mining}}},
  author = {Wilson, Garrett and Doppa, Janardhan Rao and Cook, Diane J.},
  date = {2020-08-20},
  series = {{{KDD}} '20},
  pages = {1768--1778},
  publisher = {Association for Computing Machinery},
  location = {New York, NY, USA},
  doi = {10.1145/3394486.3403228},
  url = {https://dl.acm.org/doi/10.1145/3394486.3403228},
  abstract = {Domain adaptation (DA) offers a valuable means to reuse data and models for new problem domains. However, robust techniques have not yet been considered for time series data with varying amounts of data availability. In this paper, we make three main contributions to fill this gap. First, we propose a novel Convolutional deep Domain Adaptation model for Time Series data (CoDATS) that significantly improves accuracy and training time over state-of-the-art DA strategies on real-world sensor data benchmarks. By utilizing data from multiple source domains, we increase the usefulness of CoDATS to further improve accuracy over prior single-source methods, particularly on complex time series datasets that have high variability between domains. Second, we propose a novel Domain Adaptation with Weak Supervision (DA-WS) method by utilizing weak supervision in the form of target-domain label distributions, which may be easier to collect than additional data labels. Third, we perform comprehensive experiments on diverse real-world datasets to evaluate the effectiveness of our domain adaptation and weak supervision methods. Results show that CoDATS for single-source DA significantly improves over the state-of-the-art methods, and we achieve additional improvements in accuracy using data from multiple source domains and weakly supervised signals.},
  isbn = {978-1-4503-7998-4},
  keywords = {domain adaptation,human activity recognition,time series,transfer learning,weak supervision},
  file = {C:\Users\sjelic\Zotero\storage\FS3U3X2K\Wilson et al. - 2020 - Multi-Source Deep Domain Adaptation with Weak Supe.pdf}
}

@inproceedings{wilsonMultiSourceDeepDomain2020a,
  title = {Multi-{{Source Deep Domain Adaptation}} with {{Weak Supervision}} for {{Time-Series Sensor Data}}},
  booktitle = {Proceedings of the 26th {{ACM SIGKDD International Conference}} on {{Knowledge Discovery}} \& {{Data Mining}}},
  author = {Wilson, Garrett and Doppa, Janardhan Rao and Cook, Diane J.},
  date = {2020-08-20},
  series = {{{KDD}} '20},
  pages = {1768--1778},
  publisher = {Association for Computing Machinery},
  location = {New York, NY, USA},
  doi = {10.1145/3394486.3403228},
  url = {https://dl.acm.org/doi/10.1145/3394486.3403228},
  abstract = {Domain adaptation (DA) offers a valuable means to reuse data and models for new problem domains. However, robust techniques have not yet been considered for time series data with varying amounts of data availability. In this paper, we make three main contributions to fill this gap. First, we propose a novel Convolutional deep Domain Adaptation model for Time Series data (CoDATS) that significantly improves accuracy and training time over state-of-the-art DA strategies on real-world sensor data benchmarks. By utilizing data from multiple source domains, we increase the usefulness of CoDATS to further improve accuracy over prior single-source methods, particularly on complex time series datasets that have high variability between domains. Second, we propose a novel Domain Adaptation with Weak Supervision (DA-WS) method by utilizing weak supervision in the form of target-domain label distributions, which may be easier to collect than additional data labels. Third, we perform comprehensive experiments on diverse real-world datasets to evaluate the effectiveness of our domain adaptation and weak supervision methods. Results show that CoDATS for single-source DA significantly improves over the state-of-the-art methods, and we achieve additional improvements in accuracy using data from multiple source domains and weakly supervised signals.},
  isbn = {978-1-4503-7998-4},
  file = {C:\Users\sjelic\Zotero\storage\8YI6JGKK\Wilson et al. - 2020 - Multi-Source Deep Domain Adaptation with Weak Supervision for Time-Series Sensor Data.pdf}
}

@article{wilsonSurveyUnsupervisedDeep2020,
  title = {A {{Survey}} of {{Unsupervised Deep Domain Adaptation}}},
  author = {Wilson, Garrett and Cook, Diane J.},
  date = {2020-07-05},
  journaltitle = {ACM Transactions on Intelligent Systems and Technology},
  shortjournal = {ACM Trans. Intell. Syst. Technol.},
  volume = {11},
  number = {5},
  pages = {51:1--51:46},
  issn = {2157-6904},
  doi = {10.1145/3400066},
  url = {https://doi.org/10.1145/3400066},
  abstract = {Deep learning has produced state-of-the-art results for a variety of tasks. While such approaches for supervised learning have performed well, they assume that training and testing data are drawn from the same distribution, which may not always be the case. As a complement to this challenge, single-source unsupervised domain adaptation can handle situations where a network is trained on labeled data from a source domain and unlabeled data from a related but different target domain with the goal of performing well at test-time on the target domain. Many single-source and typically homogeneous unsupervised deep domain adaptation approaches have thus been developed, combining the powerful, hierarchical representations from deep learning with domain adaptation to reduce reliance on potentially costly target data labels. This survey will compare these approaches by examining alternative methods, the unique and common elements, results, and theoretical insights. We follow this with a look at application areas and open research directions.},
  keywords = {deep learning,Domain adaptation,generative adversarial networks},
  file = {C:\Users\sjelic\Zotero\storage\WB32CB4T\Wilson and Cook - 2020 - A Survey of Unsupervised Deep Domain Adaptation.pdf}
}

@article{wilsonSurveyUnsupervisedDeep2020a,
  title = {A {{Survey}} of {{Unsupervised Deep Domain Adaptation}}},
  author = {Wilson, Garrett and Cook, Diane J.},
  date = {2020-07-05},
  journaltitle = {ACM Trans. Intell. Syst. Technol.},
  volume = {11},
  number = {5},
  pages = {51:1--51:46},
  issn = {2157-6904},
  doi = {10.1145/3400066},
  url = {https://dl.acm.org/doi/10.1145/3400066},
  abstract = {Deep learning has produced state-of-the-art results for a variety of tasks. While such approaches for supervised learning have performed well, they assume that training and testing data are drawn from the same distribution, which may not always be the case. As a complement to this challenge, single-source unsupervised domain adaptation can handle situations where a network is trained on labeled data from a source domain and unlabeled data from a related but different target domain with the goal of performing well at test-time on the target domain. Many single-source and typically homogeneous unsupervised deep domain adaptation approaches have thus been developed, combining the powerful, hierarchical representations from deep learning with domain adaptation to reduce reliance on potentially costly target data labels. This survey will compare these approaches by examining alternative methods, the unique and common elements, results, and theoretical insights. We follow this with a look at application areas and open research directions.},
  file = {C:\Users\sjelic\Zotero\storage\73BLW2ZS\Wilson and Cook - 2020 - A Survey of Unsupervised Deep Domain Adaptation.pdf}
}

@article{wuIteratedLocalSearch2024,
  title = {An {{Iterated Local Search Heuristic}} for the {{Multi-Trip Vehicle Routing Problem}} with {{Multiple Time Windows}}},
  author = {Wu, Yinghui and Du, Haoran and Song, Huixin},
  date = {2024-01},
  journaltitle = {Mathematics},
  volume = {12},
  number = {11},
  pages = {1712},
  publisher = {Multidisciplinary Digital Publishing Institute},
  issn = {2227-7390},
  doi = {10.3390/math12111712},
  url = {https://www.mdpi.com/2227-7390/12/11/1712},
  abstract = {This paper studies the multi-trip vehicle routing problem with multiple time windows, which extends the multi-trip vehicle routing problem by deciding not only the sequence of customers that each vehicle serves but also the service time window of each customer. It also requires that the delivery service time is within the selected time windows and that the total demand of the customers served by the vehicle on each trip does not exceed the maximum carrying capacity. For solving the studied problem, we develop a mixed integer linear programming model with the objective of minimizing the total travel distance of vehicles and design a tailored iterative local search heuristic. Within the framework of the iterative local search, an improved Solomon greedy insertion algorithm suitable for multiple time windows and multi-trip scenarios is designed to generate the initial solution, and local search operators such as Or-opt and Relocate, as well as Random Exchange perturbation operations, are also developed. The experiment results demonstrate the effectiveness of the proposed model and algorithm and confirm that by providing customers with multiple time windows option, carriers can flexibly plan vehicle routes and select appropriate service time windows, thereby reducing the number of vehicles used and the total distance travelled and improve delivery success.},
  issue = {11},
  langid = {english},
  keywords = {iterated local search,mixed integer programming,multi-trip vehicle routing problem,multiple time windows},
  file = {C:\Users\sjelic\Zotero\storage\KJLFCYKS\Wu et al. - 2024 - An Iterated Local Search Heuristic for the Multi-Trip Vehicle Routing Problem with Multiple Time Win.pdf}
}

@online{wuSpatialAggregationTemporal2021,
  title = {Spatial {{Aggregation}} and {{Temporal Convolution Networks}} for {{Real-time Kriging}}},
  author = {Wu, Yuankai and Zhuang, Dingyi and Lei, Mengying and Labbe, Aurelie and Sun, Lijun},
  date = {2021-09-24},
  eprint = {2109.12144},
  eprinttype = {arXiv},
  eprintclass = {cs},
  doi = {10.48550/arXiv.2109.12144},
  url = {http://arxiv.org/abs/2109.12144},
  abstract = {Spatiotemporal kriging is an important application in spatiotemporal data analysis, aiming to recover/interpolate signals for unsampled/unobserved locations based on observed signals. The principle challenge for spatiotemporal kriging is how to effectively model and leverage the spatiotemporal dependencies within the data. Recently, graph neural networks (GNNs) have shown great promise for spatiotemporal kriging tasks. However, standard GNNs often require a carefully designed adjacency matrix and specific aggregation functions, which are inflexible for general applications/problems. To address this issue, we present SATCN -- Spatial Aggregation and Temporal Convolution Networks -- a universal and flexible framework to perform spatiotemporal kriging for various spatiotemporal datasets without the need for model specification. Specifically, we propose a novel spatial aggregation network (SAN) inspired by Principal Neighborhood Aggregation, which uses multiple aggregation functions to help one node gather diverse information from its neighbors. To exclude information from unsampled nodes, a masking strategy that prevents the unsampled sensors from sending messages to their neighborhood is introduced to SAN. We capture temporal dependencies by the temporal convolutional networks, which allows our model to cope with data of diverse sizes. To make SATCN generalizable to unseen nodes and even unseen graph structures, we employ an inductive strategy to train SATCN. We conduct extensive experiments on three real-world spatiotemporal datasets, including traffic speed and climate recordings. Our results demonstrate the superiority of SATCN over traditional and GNN-based kriging models.},
  pubstate = {prepublished},
  keywords = {Computer Science - Machine Learning},
  file = {C\:\\Users\\sjelic\\Zotero\\storage\\LECDTGGK\\Wu et al. - 2021 - Spatial Aggregation and Temporal Convolution Netwo.pdf;C\:\\Users\\sjelic\\Zotero\\storage\\FPBWWFR4\\2109.html}
}

@article{xuKernelLeastAbsolute2013,
  title = {Kernel Least Absolute Shrinkage and Selection Operator Regression Classifier for Pattern Classification},
  author = {Xu, Jie and Yin, Jun},
  date = {2013},
  journaltitle = {IET Computer Vision},
  volume = {7},
  number = {1},
  pages = {48--55},
  issn = {1751-9640},
  doi = {10.1049/iet-cvi.2011.0193},
  url = {https://onlinelibrary.wiley.com/doi/abs/10.1049/iet-cvi.2011.0193},
  abstract = {The feature vectors in feature space are more likely to be linearly separable than the observations in input space. To enhance the separability of the feature vectors, the authors perform least absolute shrinkage and selection operator (LASSO) regression in the reproducing kernel Hilbert space and develop a kernel LASSO regression classifier (LASSO-KRC). Based on the theory of calculus, least squares optimisation with L1-norm regularised constraints can be reformulated into another equivalent form. Without an explicit mapping function, the solution to the optimisation problem can be obtained by solving a convex optimisation problem with any symmetric kernel function. LASSO-KRC is applied to pattern classification and appears to outperform nearest neighbour classifier, minimum distance classifier, sparse representation classifier and linear regression classifier.},
  langid = {english},
  keywords = {feature vectors,Hilbert spaces,kernel Hilbert space,kernel LASSO regression classifier,kernel least absolute shrinkage,LASSO-KRC,least squares approximations,least squares optimisation,linear regression classifier,optimisation,pattern classification,regression analysis,selection operator regression classifier,vectors},
  file = {C\:\\Users\\sjelic\\Zotero\\storage\\JWTFLRXM\\Xu and Yin - 2013 - Kernel least absolute shrinkage and selection oper.pdf;C\:\\Users\\sjelic\\Zotero\\storage\\YQMU6K5D\\iet-cvi.2011.html}
}

@article{yangApproximatingProbabilisticGroup2022,
  title = {Approximating Probabilistic Group Steiner Trees in Graphs},
  author = {Yang, Shuang and Sun, Yahui and Liu, Jiesong and Xiao, Xiaokui and Li, Rong-Hua and Wei, Zhewei},
  date = {2022-10-01},
  journaltitle = {Proceedings of the VLDB Endowment},
  shortjournal = {Proc. VLDB Endow.},
  volume = {16},
  number = {2},
  pages = {343--355},
  issn = {2150-8097},
  doi = {10.14778/3565816.3565834},
  url = {https://doi.org/10.14778/3565816.3565834},
  abstract = {Consider an edge-weighted graph, and a number of properties of interests (PoIs). Each vertex has a probability of exhibiting each PoI. The joint probability that a set of vertices exhibits a PoI is the probability that this set contains at least one vertex that exhibits this PoI. The probabilistic group Steiner tree problem is to find a tree such that (i) for each PoI, the joint probability that the set of vertices in this tree exhibits this PoI is no smaller than a threshold value, e.g., 0.97; and (ii) the total weight of edges in this tree is the minimum. Solving this problem is useful for mining various graphs with uncertain vertex properties, but is NP-hard. The existing work focuses on certain cases, and cannot perform this task. To meet this challenge, we propose 3 approximation algorithms for solving the above problem. Let |Γ| be the number of PoIs, and ξ be an upper bound of the number of vertices for satisfying the threshold value of exhibiting each PoI. Algorithms 1 and 2 have tight approximation guarantees proportional to |Γ| and ξ, and exponential time complexities with respect to ξ and |Γ|, respectively. In comparison, Algorithm 3 has a looser approximation guarantee proportional to, and a polynomial time complexity with respect to, both |Γ| and ξ. Experiments on real and large datasets show that the proposed algorithms considerably outperform the state-of-the-art related work for finding probabilistic group Steiner trees in various cases.}
}

@article{yKernelizedElasticNet2016,
  title = {Kernelized {{Elastic Net Regularization}}: {{Generalization Bounds}}, and {{Sparse Recovery}}},
  shorttitle = {Kernelized {{Elastic Net Regularization}}},
  author = {Y, Feng and Sg, Lv and H, Hang and Ja, Suykens},
  date = {2016-03},
  journaltitle = {Neural computation},
  volume = {28},
  number = {3},
  eprint = {26735744},
  eprinttype = {pubmed},
  publisher = {Neural Comput},
  issn = {1530-888X},
  doi = {10.1162/NECO_a_00812},
  url = {https://pubmed.ncbi.nlm.nih.gov/26735744/},
  abstract = {Kernelized elastic net regularization (KENReg) is a kernelization of the well-known elastic net regularization (Zou \& Hastie, 2005). The kernel in KENReg is not required to be a Mercer kernel since it learns from a kernelized dictionary in the coefficient space. Feng, Yang, Zhao, Lv, and Suykens …},
  langid = {english},
  file = {C:\Users\sjelic\Zotero\storage\XMYNAD9X\26735744.html}
}

@article{zachariasenRectilinearGroupSteiner2003,
  title = {Rectilinear Group {{Steiner}} Trees and Applications in {{VLSI}} Design},
  author = {Zachariasen, Martin and Rohe, André},
  date = {2003-01-01},
  journaltitle = {Mathematical Programming},
  shortjournal = {Math. Program., Ser. B},
  volume = {94},
  number = {2},
  pages = {407--433},
  issn = {1436-4646},
  doi = {10.1007/s10107-002-0326-x},
  url = {https://doi.org/10.1007/s10107-002-0326-x},
  abstract = {Given a set of disjoint groups of points in the plane, the rectilinear group Steiner tree problem is the problem of finding a shortest interconnection (under the rectilinear metric) which includes at least one point from each group. This is an important generalization of the well-known rectilinear Steiner tree problem which has direct applications in VLSI design: in the detailed routing phase the logical units typically allow the nets to connect to several electrically equivalent ports. We present a first (tailored) exact algorithm for solving the rectilinear group Steiner tree problem (and related variants of the problem). The algorithm essentially constructs a subgraph of the corresponding Hanan grid on which existing algorithms for solving the Steiner tree problem in graphs are applied. The reductions of the Hanan grid are performed by applying point deletions and by generating full Steiner trees on the remaining points. Experimental results for real-world VLSI instances with up to 100 groups are presented.},
  langid = {english},
  keywords = {Direct Application,Exact Algorithm,Logical Unit,Related Variant,Steiner Tree}
}

@article{zachariasenRectilinearGroupSteiner2003a,
  title = {Rectilinear Group {{Steiner}} Trees and Applications in {{VLSI}} Design},
  author = {Zachariasen, Martin and Rohe, André},
  date = {2003-01-01},
  journaltitle = {Mathematical Programming},
  shortjournal = {Math. Program., Ser. B},
  volume = {94},
  number = {2},
  pages = {407--433},
  issn = {1436-4646},
  doi = {10.1007/s10107-002-0326-x},
  url = {https://doi.org/10.1007/s10107-002-0326-x},
  abstract = {Given a set of disjoint groups of points in the plane, the rectilinear group Steiner tree problem is the problem of finding a shortest interconnection (under the rectilinear metric) which includes at least one point from each group. This is an important generalization of the well-known rectilinear Steiner tree problem which has direct applications in VLSI design: in the detailed routing phase the logical units typically allow the nets to connect to several electrically equivalent ports. We present a first (tailored) exact algorithm for solving the rectilinear group Steiner tree problem (and related variants of the problem). The algorithm essentially constructs a subgraph of the corresponding Hanan grid on which existing algorithms for solving the Steiner tree problem in graphs are applied. The reductions of the Hanan grid are performed by applying point deletions and by generating full Steiner trees on the remaining points. Experimental results for real-world VLSI instances with up to 100 groups are presented.},
  langid = {english},
  keywords = {Direct Application,Exact Algorithm,Logical Unit,Related Variant,Steiner Tree}
}

@article{zelikovskySeriesApproximationAlgorithms1997,
  title = {A Series of Approximation Algorithms for the Acyclic Directed Steiner Tree Problem},
  author = {Zelikovsky, A.},
  date = {1997-05-01},
  journaltitle = {Algorithmica},
  shortjournal = {Algorithmica},
  volume = {18},
  number = {1},
  pages = {99--110},
  issn = {1432-0541},
  doi = {10.1007/BF02523690},
  url = {https://doi.org/10.1007/BF02523690},
  abstract = {Given an acyclic directed network, a subsetS of nodes (terminals), and a rootr, theacyclic directed Steiner tree problem requires a minimum-cost subnetwork which contains paths fromr to each terminal. It is known that unlessNP⊆DTIME[npolylogn] no polynomial-time algorithm can guarantee better than (lnk)/4-approximation, wherek is the number of terminals. In this paper we give anO(kε)-approximation algorithm for any ε{$>$}0. This result improves the previously knownk-approximation.},
  langid = {english},
  keywords = {Algorithms,Approximations,Steiner tree},
  file = {C:\Users\sjelic\Zotero\storage\UUFTDLGA\Zelikovsky - 1997 - A series of approximation algorithms for the acycl.pdf}
}

@article{zhangApproximationAlgorithmGroup2022,
  title = {An Approximation Algorithm for the Group Prize-Collecting {{Steiner}} Tree Problem with Submodular Penalties},
  author = {Zhang, Jiaxuan and Gao, Suogang and Hou, Bo and Liu, Wen},
  date = {2022-08-05},
  journaltitle = {Computational and Applied Mathematics},
  shortjournal = {Comp. Appl. Math.},
  volume = {41},
  number = {6},
  pages = {274},
  issn = {1807-0302},
  doi = {10.1007/s40314-022-01984-2},
  url = {https://doi.org/10.1007/s40314-022-01984-2},
  abstract = {In this paper, we consider the group prize-collecting Steiner tree problem with submodular penalties (GPCST-SP problem). In this problem, we are given an undirected connected graph \$\$G=(V,E)\$\$with a pre-specified root r and a partition \$\$\textbackslash mathcal \{V\}=\textbackslash\{V\_0,V\_1,\textbackslash ldots ,V\_k\textbackslash\}\$\$of V with \$\$r\textbackslash in V\_0\$\$. Assume \$\$c: E\textbackslash rightarrow \textbackslash mathbb \{R\}\_+\$\$is an edge cost function and \$\$p: 2\textasciicircum\{\textbackslash mathcal \{V\}\}\textbackslash rightarrow \textbackslash mathbb \{R\}\_+\$\$is a submodular penalty function, where \$\$\textbackslash mathbb \{R\}\_+\$\$is the set of nonnegative real numbers. For a group \$\$V\_i\textbackslash in \textbackslash mathcal \{V\}\$\$, we say it is spanned by a tree if the tree contains at least one vertex of that group. The goal of the GPCST-SP problem is to find an r-rooted tree that minimizes the costs of the edges in the tree plus the penalty cost of the subcollection \$\$\textbackslash mathcal \{S\}\$\$containing these groups not spanned by the tree. Our main result is a 2I-approximation algorithm for the problem, where \$\$I=\textbackslash max \textbackslash\{ |V\_i| \textbackslash mid i=0,1,2,\textbackslash ldots ,k\textbackslash\}\$\$.},
  langid = {english},
  keywords = {90C27,90C35,Approximation algorithm,Group prize-collecting Steiner tree problem,Submodular function},
  file = {C:\Users\sjelic\Zotero\storage\GM84ZXC5\Zhang et al. - 2022 - An approximation algorithm for the group prize-col.pdf}
}

@article{zhangPostmodelselectionInferenceLinear2022,
  title = {Post-Model-Selection Inference in Linear Regression Models: {{An}} Integrated Review},
  shorttitle = {Post-Model-Selection Inference in Linear Regression Models},
  author = {Zhang, Dongliang and Khalili, Abbas and Asgharian, Masoud},
  date = {2022-01-01},
  journaltitle = {Statistics Surveys},
  shortjournal = {Statist. Surv.},
  volume = {16},
  issn = {1935-7516},
  doi = {10.1214/22-SS135},
  url = {https://projecteuclid.org/journals/statistics-surveys/volume-16/issue-none/Post-model-selection-inference-in-linear-regression-models--An/10.1214/22-SS135.full},
  abstract = {The research on statistical inference after data-driven model selection can be traced as far back as Koopmans (1949). The intensive research on modern model selection methods for high-dimensional data over the past three decades revived the interest in statistical inference after model selection. In recent years, there has been a surge of articles on statistical inference after model selection and now a rather vast literature exists on this topic. Our manuscript aims at presenting a holistic review of post-model-selection inference in linear regression models, while also incorporating perspectives from high-dimensional inference in these models. We first give a simulated example motivating the necessity for valid statistical inference after model selection. We then provide theoretical insights explaining the phenomena observed in the example. This is done through a literature survey on the post-selection sampling distribution of regression parameter estimators and properties of coverage probabilities of na¨ıve confidence intervals. Categorized according to two types of estimation targets, namely the population- and projection-based regression coefficients, we present a review of recent uncertainty assessment methods. We also discuss possible pros and cons for the confidence intervals constructed by different methods.},
  issue = {none},
  langid = {english},
  file = {C\:\\Users\\sjelic\\Zotero\\storage\\AU2HWSF9\\Zhang et al. - 2022 - Post-model-selection inference in linear regressio.pdf;C\:\\Users\\sjelic\\Zotero\\storage\\AYXZ8H8S\\ss135supp.pdf}
}

@article{zhangPrioritizationCancerDriver2022,
  title = {Prioritization of Cancer Driver Gene with Prize-Collecting Steiner Tree by Introducing an Edge Weighted Strategy in the Personalized Gene Interaction Network},
  author = {Zhang, Shao-Wu and Wang, Zhen-Nan and Li, Yan and Guo, Wei-Feng},
  date = {2022-08-16},
  journaltitle = {BMC Bioinformatics},
  shortjournal = {BMC Bioinformatics},
  volume = {23},
  number = {1},
  pages = {341},
  issn = {1471-2105},
  doi = {10.1186/s12859-022-04802-y},
  url = {https://doi.org/10.1186/s12859-022-04802-y},
  abstract = {Cancer is a heterogeneous disease in which tumor genes cooperate as well as adapt and evolve to the changing conditions for individual patients. It is a meaningful task to discover the personalized cancer driver genes that can provide diagnosis and target drug for individual patients. However, most of existing methods mainly ranks potential personalized cancer driver genes by considering the patient-specific nodes information on the gene/protein interaction network. These methods ignore the personalized edge weight information in gene interaction network, leading to false positive results.},
  keywords = {Driver gene,Gene interaction network,Personalized cancer,Prize-collecting Steiner tree},
  file = {C:\Users\sjelic\Zotero\storage\VV3MITMR\Zhang et al. - 2022 - Prioritization of cancer driver gene with prize-co.pdf}
}

@inproceedings{zhaoLearningInvariantRepresentations2019,
  title = {On {{Learning Invariant Representations}} for {{Domain Adaptation}}},
  booktitle = {Proceedings of the 36th {{International Conference}} on {{Machine Learning}}},
  author = {Zhao, Han and Combes, Remi Tachet Des and Zhang, Kun and Gordon, Geoffrey},
  date = {2019-05-24},
  pages = {7523--7532},
  publisher = {PMLR},
  issn = {2640-3498},
  url = {https://proceedings.mlr.press/v97/zhao19a.html},
  abstract = {Due to the ability of deep neural nets to learn rich representations, recent advances in unsupervised domain adaptation have focused on learning domain-invariant features that achieve a small error on the source domain. The hope is that the learnt representation, together with the hypothesis learnt from the source domain, can generalize to the target domain. In this paper, we first construct a simple counterexample showing that, contrary to common belief, the above conditions are not sufficient to guarantee successful domain adaptation. In particular, the counterexample exhibits conditional shift: the class-conditional distributions of input features change between source and target domains. To give a sufficient condition for domain adaptation, we propose a natural and interpretable generalization upper bound that explicitly takes into account the aforementioned shift. Moreover, we shed new light on the problem by proving an information-theoretic lower bound on the joint error of any domain adaptation method that attempts to learn invariant representations. Our result characterizes a fundamental tradeoff between learning invariant representations and achieving small joint error on both domains when the marginal label distributions differ from source to target. Finally, we conduct experiments on real-world datasets that corroborate our theoretical findings. We believe these insights are helpful in guiding the future design of domain adaptation and representation learning algorithms.},
  eventtitle = {International {{Conference}} on {{Machine Learning}}},
  langid = {english},
  file = {C\:\\Users\\sjelic\\Zotero\\storage\\8JGXYXKW\\Zhao et al. - 2019 - On Learning Invariant Representations for Domain A.pdf;C\:\\Users\\sjelic\\Zotero\\storage\\IMKRQXW5\\Zhao et al. - 2019 - On Learning Invariant Representations for Domain A.pdf}
}

@article{zhengOptimizingFeatureSelection2024,
  title = {Optimizing Feature Selection with Gradient Boosting Machines in {{PLS}} Regression for Predicting Moisture and Protein in Multi-Country Corn Kernels via {{NIR}} Spectroscopy},
  author = {Zheng, Runyu and Jia, Yuyao and Ullagaddi, Chidanand and Allen, Cody and Rausch, Kent and Singh, Vijay and Schnable, James C. and Kamruzzaman, Mohammed},
  date = {2024-10-30},
  journaltitle = {Food Chemistry},
  shortjournal = {Food Chemistry},
  volume = {456},
  pages = {140062},
  issn = {0308-8146},
  doi = {10.1016/j.foodchem.2024.140062},
  url = {https://www.sciencedirect.com/science/article/pii/S0308814624017126},
  abstract = {Differences in moisture and protein content impact both nutritional value and processing efficiency of corn kernels. Near-infrared (NIR) spectroscopy can be used to estimate kernel composition, but models trained on a few environments may underestimate error rates and bias. We assembled corn samples from diverse international environments and used NIR with chemometrics and partial least squares regression (PLSR) to determine moisture and protein. The potential of five feature selection methods to improve prediction accuracy was assessed by extracting sensitive wavelengths. Gradient boosting machines (GBMs), particularly CatBoost and LightGBM, were found to effectively select crucial wavelengths for moisture (1409, 1900, 1908, 1932, 1953, 2174~nm) and protein (887, 1212, 1705, 1891, 2097, 2456~nm). SHAP plots highlighted significant wavelength contributions to model prediction. These results illustrate GBMs' effectiveness in feature engineering for agricultural and food sector applications, including developing multi-country global calibration models for moisture and protein in corn kernels.},
  keywords = {Component prediction,Corn kernels,Feature selection,Gradient boosting machine (GBM),Near-infrared (NIR) spectroscopy,Partial least squares regression (PLSR),SHapley additive exPlanations (SHAP)},
  file = {C\:\\Users\\sjelic\\Zotero\\storage\\XHDW68RE\\Zheng et al. - 2024 - Optimizing feature selection with gradient boostin.pdf;C\:\\Users\\sjelic\\Zotero\\storage\\M59JN283\\S0308814624017126.html}
}

@article{zhengRemoteSensingCrop2014,
  title = {Remote Sensing of Crop Residue and Tillage Practices: {{Present}} Capabilities and Future Prospects},
  shorttitle = {Remote Sensing of Crop Residue and Tillage Practices},
  author = {Zheng, Baojuan and Campbell, James B. and Serbin, Guy and Galbraith, John M.},
  date = {2014-05},
  journaltitle = {Soil and Tillage Research},
  shortjournal = {Soil and Tillage Research},
  volume = {138},
  pages = {26--34},
  issn = {01671987},
  doi = {10.1016/j.still.2013.12.009},
  url = {https://linkinghub.elsevier.com/retrieve/pii/S0167198714000051},
  langid = {english},
  file = {C:\Users\sjelic\Zotero\storage\4K44J2EI\Zheng et al. - 2014 - Remote sensing of crop residue and tillage practic.pdf}
}

@article{zhenMultidepotMultitripVehicle2020,
  title = {Multi-Depot Multi-Trip Vehicle Routing Problem with Time Windows and Release Dates},
  author = {Zhen, Lu and Ma, Chengle and Wang, Kai and Xiao, Liyang and Zhang, Wei},
  date = {2020-03-01},
  journaltitle = {Transportation Research Part E: Logistics and Transportation Review},
  shortjournal = {Transportation Research Part E: Logistics and Transportation Review},
  volume = {135},
  pages = {101866},
  issn = {1366-5545},
  doi = {10.1016/j.tre.2020.101866},
  url = {https://www.sciencedirect.com/science/article/pii/S1366554519301486},
  abstract = {This study investigates a multi-depot multi-trip vehicle routing problem with time windows and release dates, which is a practical problem in the last mile distribution operations. This problem aims to design a set of trips for the fleet of vehicles supplied by different depots for minimizing total traveling time. It addresses some realistic considerations, such as the customers’ time windows and the release date of customers’ packages. The problem is formulated as a mixed integer programming model. A hybrid particle swarm optimization algorithm and a hybrid genetic algorithm are developed to solve this problem. Extensive numerical experiments are conducted to validate the effectiveness of the proposed model and the efficiency of the proposed solution methods. The experimental results show that our proposed algorithms can obtain near-optimal solutions for small-scale problem instances, and solve some large-scale instances with up to 200 customers, 20 depots and 40 vehicles in reasonable computation time.},
  keywords = {Multi-depot,Multi-trip,Release date,Time window,Vehicle routing},
  file = {C\:\\Users\\sjelic\\Zotero\\storage\\6ASKT47Q\\Zhen и сар. - 2020 - Multi-depot multi-trip vehicle routing problem with time windows and release dates.pdf;C\:\\Users\\sjelic\\Zotero\\storage\\STX63T49\\S1366554519301486.html}
}

@article{zouFusionConvolutionalNeural2025,
  title = {Fusion of Convolutional Neural Network with {{XGBoost}} Feature Extraction for Predicting Multi-Constituents in Corn Using near Infrared Spectroscopy},
  author = {Zou, Xin and Wang, Qiaoyun and Chen, Yinji and Wang, Jilong and Xu, Shunyuan and Zhu, Ziheng and Yan, Chongyue and Shan, Peng and Wang, Shuyu and Fu, YongQing},
  date = {2025-01-15},
  journaltitle = {Food Chemistry},
  shortjournal = {Food Chemistry},
  volume = {463},
  pages = {141053},
  issn = {0308-8146},
  doi = {10.1016/j.foodchem.2024.141053},
  url = {https://www.sciencedirect.com/science/article/pii/S0308814624027031},
  abstract = {Near-infrared (NIR) spectroscopy has been widely utilized to predict multi-constituents of corn in agriculture. However, directly extracting constituent information from the NIR spectra is challenging due to many issues such as broad absorption band, overlapping and non-specific nature. To solve these problems and extract implicit features from the raw data of NIR spectra to improve performance of quantitative models, a one-dimensional shallow convolutional neural network (CNN) model based on an eXtreme Gradient Boosting (XGBoost) feature extraction method was proposed in this paper. The leaf node feature information in the XGBoost was encoded and reconstructed to obtain the implicit features of raw data in the NIR spectra. A two-parametric Swish (TSwish or TS) activation function was proposed to improve the performance of CNN, and the elastic net (EN) was also applied to avoid the overfitting problem of the CNN model. Performance of the developed XGBoost-CNN-TS-EN model was evaluated using two public NIR spectroscopy datasets of corn and soil, and the obtained determination coefficients (R2) for moisture, oil, protein, and starch of the corn on test set were 0.993, 0.991, 0.998, and 0.992, respectively, with that of the soil organic matter being 0.992. The XGBoost-CNN-TS-EN model exhibits superior stability, good prediction accuracy, and generalization ability, demonstrating its great potentials for quantitative analysis of multi-constituents in spectroscopic applications.},
  keywords = {Activation function,Convolutional neural network,Elastic net,Near-infrared spectroscopy,XGBoost feature extraction},
  file = {C\:\\Users\\sjelic\\Zotero\\storage\\ZX9JGLMG\\Zou et al. - 2025 - Fusion of convolutional neural network with XGBoos.pdf;C\:\\Users\\sjelic\\Zotero\\storage\\XZCPS8TW\\S0308814624027031.html}
}

@article{zouRegularizationVariableSelection2005,
  title = {Regularization and Variable Selection via the Elastic Net},
  author = {Zou, Hui and Hastie, Trevor},
  date = {2005-04},
  journaltitle = {Journal of the Royal Statistical Society: Series B (Statistical Methodology)},
  shortjournal = {J Royal Statistical Soc B},
  volume = {67},
  number = {2},
  pages = {301--320},
  issn = {1369-7412, 1467-9868},
  doi = {10.1111/j.1467-9868.2005.00503.x},
  url = {https://onlinelibrary.wiley.com/doi/10.1111/j.1467-9868.2005.00503.x},
  abstract = {We propose the elastic net, a new regularization and variable selection method. Real world data and a simulation study show that the elastic net often outperforms the lasso, while enjoying a similar sparsity of representation. In addition, the elastic net encourages a grouping effect, where strongly correlated predictors tend to be in or out of the model together. The elastic net is particularly useful when the number of predictors (p) is much bigger than the number of observations (n). By contrast, the lasso is not a very satisfactory variable selection method in the p n case. An algorithm called LARS-EN is proposed for computing elastic net regularization paths efficiently, much like algorithm LARS does for the lasso.},
  langid = {english},
  file = {C:\Users\sjelic\Zotero\storage\ZLC5D4AW\Zou and Hastie - 2005 - Regularization and variable selection via the elas.pdf}
}