Article
Automation & Control Systems
Huangke Chen, Ran Cheng, Witold Pedrycz, Yaochu Jin
Summary: This paper proposes a method to solve multiobjective optimization problems through multi-stage evolutionary search, highlighting convergence and diversity in different search stages. The algorithm balances and addresses the issues in multiobjective optimization through two stages.
IEEE TRANSACTIONS ON SYSTEMS MAN CYBERNETICS-SYSTEMS
(2021)
Article
Automation & Control Systems
Haokai Hong, Min Jiang, Gary G. Yen
Summary: The large-scale multiobjective optimization problem (LSMOP) involves optimizing multiple conflicting objectives and hundreds of decision variables. Existing algorithms often focus on improving performance but pay little attention to improving insensitivity. We propose an evolutionary algorithm based on Monte Carlo tree search to improve the performance and insensitivity of solving LSMOPs.
IEEE TRANSACTIONS ON CYBERNETICS
(2023)
Review
Computer Science, Artificial Intelligence
Tingyang Wei, Shibin Wang, Jinghui Zhong, Dong Liu, Jun Zhang
Summary: This paper presents a detailed exposition on the research in the field of evolutionary multitask optimization (EMTO), revealing the core components of EMTO algorithms and the fusion between EMTO and traditional evolutionary algorithms. By analyzing the associations of different strategies in various branches of EMTO, this review uncovers research trends and potentially important directions, as well as mentions interesting real-world applications.
IEEE TRANSACTIONS ON EVOLUTIONARY COMPUTATION
(2022)
Article
Computer Science, Artificial Intelligence
Jing Liang, Xuanxuan Ban, Kunjie Yu, Boyang Qu, Kangjia Qiao, Caitong Yue, Ke Chen, Kay Chen Tan
Summary: Handling constrained multiobjective optimization problems is challenging due to the need to simultaneously optimize multiple conflicting objectives subject to various constraints. This article provides a comprehensive survey of evolutionary constrained multiobjective optimization. It categorizes and analyzes numerous algorithms, reviews benchmark test problems, investigates the performance of constraint handling techniques and algorithms, discusses emerging and representative applications, and outlines future research directions.
IEEE TRANSACTIONS ON EVOLUTIONARY COMPUTATION
(2023)
Article
Computer Science, Artificial Intelligence
Wenhua Li, Tao Zhang, Rui Wang, Hisao Ishibuchi
Summary: This study proposes an MMEA-WI algorithm based on a weighted indicator for solving multimodal multiobjective problems, which outperforms some state-of-the-art MMEAs in terms of performance metrics. By integrating diversity information and introducing a convergence archive, the algorithm effectively maintains diversity and ensures a better approximation of the Pareto-optimal front.
IEEE TRANSACTIONS ON EVOLUTIONARY COMPUTATION
(2021)
Article
Computer Science, Artificial Intelligence
Genghui Li, Qingfu Zhang, Zhenkun Wang
Summary: This article introduces a special multitasking optimization problem, called the competitive MTOP (CMTOP), where all tasks' objectives are comparable and the optimal solution is the best among all individual problems. An evolutionary algorithm with online resource allocation strategy and adaptive information transfer mechanism is proposed to solve the CMTOP. Experimental results on benchmark and real-world problems demonstrate the effectiveness and efficiency of the proposed algorithm.
IEEE TRANSACTIONS ON EVOLUTIONARY COMPUTATION
(2022)
Article
Computer Science, Artificial Intelligence
Mengjun Ming, Rui Wang, Hisao Ishibuchi, Tao Zhang
Summary: This article proposes a novel method, DD-CMOEA, that addresses the issue of effectively exploring and exploiting infeasible solutions in constrained multiobjective optimization problems. The method features dual stages and dual populations, and has been shown to outperform five state-of-the-art CMOEAs in extensive experiments.
IEEE TRANSACTIONS ON EVOLUTIONARY COMPUTATION
(2022)
Article
Computer Science, Artificial Intelligence
Mengjun Ming, Anupam Trivedi, Rui Wang, Dipti Srinivasan, Tao Zhang
Summary: The study introduces a dual-population-based evolutionary algorithm, c-DPEA, for constrained multiobjective optimization problems (CMOPs), which achieves a balance between convergence and diversity through the design of novel self-adaptive penalty and fitness functions. Extensive experiments demonstrate the superiority of c-DPEA over six state-of-the-art CMOEAs on most test problems.
IEEE TRANSACTIONS ON EVOLUTIONARY COMPUTATION
(2021)
Article
Automation & Control Systems
Cheng He, Ran Cheng, Danial Yazdani
Summary: In large-scale multiobjective optimization, the proposed adaptive offspring generation method effectively generates promising candidate solutions, enhancing convergence and maintaining diversity.
IEEE TRANSACTIONS ON SYSTEMS MAN CYBERNETICS-SYSTEMS
(2022)
Article
Automation & Control Systems
Ye Tian, Cheng He, Ran Cheng, Xingyi Zhang
Summary: This article provides a detailed explanation of existing diversity preservation approaches in MOEAs and their limitations, and proposes a multistage MOEA to address these limitations for better diversity performance.
IEEE TRANSACTIONS ON SYSTEMS MAN CYBERNETICS-SYSTEMS
(2021)
Article
Engineering, Electrical & Electronic
Zi He, Yu-Sheng Li, Xiao Huang, Chen-Feng Yang, Ru-Shan Chen
Summary: An efficient electromagnetic scattering solution for analyzing scattering from electrically large metallic/dielectric bodies of revolution (BoRs) is proposed, using two-dimensional non-uniform rational B-spline curves to describe the geometrical shape and establishing the relationship between local random variables and spatial basis functions. The algorithm allows for quick calculation of radar cross section for perturbed geometries, saving computational resources compared to the Monte Carlo method. The accuracy and efficiency of the proposed method is demonstrated through numerical results.
IEEE TRANSACTIONS ON ANTENNAS AND PROPAGATION
(2021)
Article
Automation & Control Systems
Yuchao Su, Qiuzhen Lin, Zhong Ming, Kay Chen Tan
Summary: This article proposes an effective method called Adapted Decomposed Directions (ADDs) for solving Multiobjective Optimization Problems (MOPs). Instead of using a single ideal or nadir point, each weight vector has its own ideal point for decomposition, and the decomposed directions are adaptively adjusted during the search process. The experimental results show that this method significantly improves the performance of three representative Multiobjective Evolutionary Algorithms (MOEAs) and outperforms seven competitive MOEAs in solving various artificial MOPs and a real-world MOP.
IEEE TRANSACTIONS ON CYBERNETICS
(2023)
Article
Computer Science, Artificial Intelligence
Jiawei Yuan, Hai-Lin Liu, Yew-Soon Ong, Zhaoshui He
Summary: To prevent the population from getting stuck in local areas and missing the constrained Pareto front fragments in constrained multiobjective optimization problems (CMOPs), this paper proposes a new constraint handling technique (CHT) based on an indicator. The CHT divides the promising areas into multiple subregions and prioritizes the removal of individuals with the worst fitness values in the densest subregions, improving the diversity of the population in the promising areas. Numerical experiments demonstrate the effectiveness of the proposed algorithm in handling different types of CMOPs, especially in problems where individuals easily appear in local infeasible areas dominating the constrained Pareto front fragments.
IEEE TRANSACTIONS ON EVOLUTIONARY COMPUTATION
(2022)
Article
Physics, Multidisciplinary
Anders Irback, Lucas Knuthson, Sandipan Mohanty, Carsten Peterson
Summary: Quantum annealing is a promising approach for obtaining approximate solutions to difficult optimization problems, and researchers have developed a novel spin representation for applying quantum annealing to lattice protein folding. The method performs well in solving long chain problems and achieves better solution quality compared to existing methods.
PHYSICAL REVIEW RESEARCH
(2022)
Article
Automation & Control Systems
Zhongwei Ma, Yong Wang, Wu Song
Summary: We propose a new constraint-handling technique based on a weighted sum of rankings using constrained dominance principle and Pareto dominance, which adaptively balances constraints and objectives in evolutionary optimization process. Experimental results show that our technique outperforms other representative constraint-handling techniques.
IEEE TRANSACTIONS ON SYSTEMS MAN CYBERNETICS-SYSTEMS
(2021)
Article
Management
Mohammad Ghaderi, Milosz Kadzinski
Summary: This paper introduces an analytical framework for estimating individuals' preferences by uncovering structural patterns that regulate general shapes of value functions and found that considering structural patterns at the population level considerably improves the predictive performance of the constructed value functions at the individual level.
OMEGA-INTERNATIONAL JOURNAL OF MANAGEMENT SCIENCE
(2021)
Article
Computer Science, Information Systems
Michal K. Tomczyk, Milosz Kadzinski
Summary: CIEMO/D is a novel co-evolutionary algorithm for interactive multiple objective optimization, aiming to find a region in the Pareto front highly relevant to the Decision Maker by biasing the evolutionary search with preference information. It co-evolves multiple subpopulations in a steady-state decomposition-based framework, where each subpopulation is driven by a different preference model. The algorithm successfully adapts to different DM's decision policies and demonstrates high competitiveness compared to other state-of-the-art interactive evolutionary hybrids utilizing DM's pairwise comparisons.
INFORMATION SCIENCES
(2021)
Article
Management
Marco Cinelli, Milosz Kadzinski, Grzegorz Miebs, Michael Gonzalez, Roman Slowinski
Summary: A new methodology for selecting Multiple Criteria Decision Analysis (MCDA) methods is introduced, implemented in the Multiple Criteria Decision Analysis Methods Selection Software (MCDA-MSS). The software provides guidance for analysts in choosing the most suitable MCDA method for a given decision problem, offering a comprehensive evaluation of over 200 MCDA methods based on problem characteristics.
EUROPEAN JOURNAL OF OPERATIONAL RESEARCH
(2022)
Article
Management
Anna Labijak-Kowalska, Milosz Kadzinski, Inga Spychala, Luis C. Dias, Javier Fiallos, Jonathan Patrick, Wojtek Michalowski, Ken Farion
Summary: This study proposes a novel variant of the value-based additive data envelopment analysis model and conducts a comprehensive robustness analysis of physicians' performance using mathematical programming and the Monte Carlo simulation. The results indicate a strong dependence of physicians' performances on the selected weight vectors, and provide a basis for identifying overall good performers, developing improvement plans, and recognizing challenging patients' complaints.
INTERNATIONAL TRANSACTIONS IN OPERATIONAL RESEARCH
(2023)
Article
Management
Zice Ru, Jiapeng Liu, Milosz Kadzinski, Xiuwu Liao
Summary: This paper proposes a novel Bayesian Ordinal Regression approach for multiple criteria choice and ranking problems. The approach utilizes an additive value function model to represent the Decision Maker's preferences in the form of pairwise comparisons. It applies the Bayesian rule to derive a posterior distribution over potential value functions and employs the Metropolis-Hastings method for summarizing the distribution and conducting robustness analysis.
EUROPEAN JOURNAL OF OPERATIONAL RESEARCH
(2022)
Article
Management
Krzysztof Martyn, Milosz Kadzinski
Summary: This study proposes preference learning algorithms for inferring the parameters of a sorting model from large sets of examples, with application in Multiple Criteria Decision Analysis. By utilizing artificial neural networks and gradient descent optimization algorithms, the study achieves high predictive accuracy.
EUROPEAN JOURNAL OF OPERATIONAL RESEARCH
(2023)
Review
Management
Milosz Kadzinski, Michal Wojcik, Krzysztof Ciomek
Summary: This article discusses the preference disaggregation setting in the context of multiple criteria ranking and choice problems. It reviews methods for constructing consistent recommendations and compares their performance in terms of reconstructing decision makers' preferences and delivering robust recommendations. The study also examines the impact of different parameterizations on the performance of these methods.
OMEGA-INTERNATIONAL JOURNAL OF MANAGEMENT SCIENCE
(2022)
Article
Management
Zice Ru, Jiapeng Liu, Milosz Kadzinski, Xiuwu Liao
Summary: This article proposes a family of probabilistic ordinal regression methods for multiple criteria sorting. It introduces Bayesian Ordinal Regression and Subjective Stochastic Ordinal Regression to derive the class assignments of alternatives based on provided preference information. The introduced approaches are evaluated through an experimental study involving real-world datasets.
EUROPEAN JOURNAL OF OPERATIONAL RESEARCH
(2023)
Article
Operations Research & Management Science
Anna Labijak-Kowalska, Milosz Kadzinski
Summary: This study proposes a ratio-based model for measuring the efficiency of decision-making units and introduces a robustness analysis framework that considers both interval and ordinal performances on inputs and outputs. The methodology takes advantage of the uncertainty related to imprecise data and considers all feasible input/output weight vectors determined by linear constraints. Methods are provided to verify the robustness of efficiency scores, efficiency preference relations, and efficiency ranks. Mathematical programming models are formulated to compute extreme, necessary, and possible results, and stochastic analysis using Monte Carlo simulations is incorporated to derive the probability distribution of different outcomes. The framework is implemented in R and made available as open-source software. Its use is illustrated through case studies on Chinese ports and industrial robots.
OPERATIONAL RESEARCH
(2023)
Article
Chemistry, Multidisciplinary
Anna Labijak-Kowalska, Milosz Kadzinski, Weronika Mrozek
Summary: We propose a new methodological framework based on additive value-based efficiency analysis, which considers inputs and outputs organized hierarchically. This approach decomposes the problem into manageable pieces and allows us to determine the strengths and weaknesses of the analyzed units. By analyzing feasible weight vectors at different hierarchy levels, we provide robust outcomes. The analysis includes three complementary perspectives: distances to the efficient unit, ranks, and pairwise preference relations. For each perspective, we determine the extreme results and the distribution of probabilistic results. We apply this method to a case study on the performance of healthcare systems in 16 Polish provinces, discussing the results based on the entire factor set and three subcategories: health improvement, efficient financial management, and consumer satisfaction. Lastly, we reveal practical conclusions derived from the hierarchical decomposition of the problem and robustness analysis.
APPLIED SCIENCES-BASEL
(2023)
Proceedings Paper
Computer Science, Artificial Intelligence
Michal K. Tomczyk, Milosz Kadzinski
Summary: The research scope of this paper is interactive evolutionary multiple objective optimization based on the preference learning paradigm. It focuses on a scenario where the Decision Maker's (DM's) aspirations evolve over time. To address the issue of computational power required to achieve satisfactory recommendations, a co-evolutionary method is proposed, which evolves sub-populations approximating the Pareto front or aligning with the DM's preferences. This method diversifies the solution set and enables quick reallocation of the preference-driven sub-population, demonstrating its usefulness in extensive experiments involving different test problems, objectives, and Decision Maker models.
PROCEEDINGS OF THE 2023 GENETIC AND EVOLUTIONARY COMPUTATION CONFERENCE, GECCO 2023
(2023)
Article
Computer Science, Interdisciplinary Applications
Jiapeng Liu, Milosz Kadzinski, Xiuwu Liaoa
Summary: We propose a preference-learning algorithm that uncovers Decision Makers' (DMs') contingent evaluation strategies in multiple criteria sorting. Our method uses holistic assignment examples derived from the analysis of performance vectors and textual descriptions. Using a mixture of threshold-based, value-driven preference models and latent topics, we construct a probabilistic model. The method automatically identifies components representing the evaluation strategies of all DMs.
INFORMS JOURNAL ON COMPUTING
(2023)
Article
Operations Research & Management Science
Jafar Rezaei, Milosz Kadzinski, Chrysoula Vana, Lori Tavasszy
Summary: This paper proposes a method to incorporate environmental evaluation criteria into supplier segmentation, analyzing the green potential of suppliers by evaluating their capabilities and willingness, and using a sorting method to solve the multi-criteria decision-making problem. It also introduces a simple method to assess the carbon footprint of the raw materials supplied by the suppliers, and combines the assessment results with the segmentation for a more useful classification.
ANNALS OF OPERATIONS RESEARCH
(2022)
Article
Computer Science, Software Engineering
Krzysztof Ciomek, Milosz Kadzinski
Summary: Polyrun is a Java library that provides methods for exploiting bounded convex polytopes, including the implementation of the Hit-and-Run algorithm. It also offers other procedures for making random steps within a polytope. The software has been used in various application areas and is free and open-source.
Article
Operations Research & Management Science
Milosz Kadzinski, Magdalena Martyn
Summary: The paper explores multi-criteria sorting problems with preference-ordered classes delimited by boundary profiles, presenting an integrated framework for modeling preference information and conducting robustness analysis. Decision makers can provide assignment examples, assignment-based pairwise comparisons, and desired class cardinalities. A diversity of recommendations can be obtained from outranking-based sorting models compatible with decision maker's preferences, quantified through six types of results.
ANNALS OF OPERATIONS RESEARCH
(2021)