Swarm and Evolutionary Computation最新文献

{"title":"Secure key based cloud security utilizing three-way protection with martino homomorphic encryption for preventing unauthorized data access","authors":"Ganji Ramanjaiah ,&nbsp;Tummala Srinivasa Ravi Kiran ,&nbsp;Ampalam Srisaila ,&nbsp;Annemneedi Lakshmanarao ,&nbsp;Komanduri Venkata Sesha Sai Ramakrishna ,&nbsp;Katakam Venkateswara Rao","doi":"10.1016/j.swevo.2025.102131","DOIUrl":"10.1016/j.swevo.2025.102131","url":null,"abstract":"<div><div>Cloud computing has transformed data storage and access by providing scalable and on-demand services. Nevertheless, it remains a priority issue to ensure the protection of sensitive data in cloud environments. Several existing security methods has fundamental shortcomings like poor threat prediction features, a failure to process encrypted data securely and high encryption time. To overcome these issues, this study proposes a novel secure key based cloud security utilizing Three-Way Protection with Martino Homomorphic Encryption for preventing unauthorized data access (SKCS-TWP-MHE-PUDA). Initially, the data are collected from Enron Email dataset. Then the input data is given to Reverse Lognormal Kalman Filter (RLKF) for data cleaning and normalization. Next, Koopman Theory Graph Convolutional Network (KTGCN) is used to analyze packet status, predict potential threats and prevent unauthorized cloud access. This real-time intrusion detection mechanism enables early anticipation of malicious activity. Meanwhile, Martino Homomorphic Encryption (MHE) is used to ensure data confidentiality by encrypting cloud-stored data such that only legitimate users decrypt and access it. The three-way security mechanism comprising user registration, intrusion detection and intrusion prevention strengthens overall protection. The performance of the proposed SKCS-TWP-MHE-PUDA method provides 26.68%, 25.75%, and 26.16% higher accuracy 29.08%, 30.70% and 16.26% higher precision when compared with existing techniques: Stochastic Gradient Descent long short-term memory dependent secure encryption approach for cloud data storage and retrieval in cloud computing environs (SGDLSTM-CDS-CCE), Blockchain Key Management: A Solution for Cloud Data Security (AES-BKY-CDS) and deep learning method with cryptographic transformation for enhancing data security in cloud environs (SqueezeNet-DS-CE) respectively.</div></div>","PeriodicalId":48682,"journal":{"name":"Swarm and Evolutionary Computation","volume":"98 ","pages":"Article 102131"},"PeriodicalIF":8.5,"publicationDate":"2025-08-25","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144893289","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Multitree genetic programming with spherical-based operators for synthetic minority over-sampling technique in unbalanced data","authors":"Yuyang Cui ,&nbsp;Ziliang Du ,&nbsp;Hongwei Ge ,&nbsp;Guangyu Zou ,&nbsp;Yaqing Hou","doi":"10.1016/j.swevo.2025.102126","DOIUrl":"10.1016/j.swevo.2025.102126","url":null,"abstract":"<div><div>Unbalanced classification is a critical challenge in machine learning, with broad applications in real-world scenarios. Recent studies have emphasized the potential of Evolutionary Computation (EC)-based approaches, particularly Genetic Programming (GP), as an effective sampling strategy for addressing class imbalance. In contrast to traditional oversampling methods that rely on neighborhood information and predefined structures, GP autonomously selects high-quality instances and evolves structures to generate new ones. However, existing GP-based approaches primarily focus on undersampling, with limited exploration of instance generation. Additionally, the traditional Single-Tree Genetic Programming (STGP) structure struggles to adapt to tasks requiring the generation of multiple candidate datasets. To address these challenges, this paper introduces MTGP-SMOTE, a novel oversampling method based on Multi-Tree Genetic Programming (MTGP). Unlike STGP, which evolves a single tree per individual, MTGP evolves multiple trees within an individual, enabling the generation of diverse new instances while evolving as a complete dataset. The method also incorporates innovative MTGP crossover and mutation operators, designed to enhance exploration by focusing on trees beyond the hemispheres of the target minority class while preserving high-quality individuals throughout the evolutionary process. Experiments on 20 unbalanced datasets demonstrate that MTGP-SMOTE significantly outperforms traditional sampling methods in reducing classifier bias and improving classification accuracy. These results underscore MTGP-SMOTE as a powerful and effective solution for addressing unbalanced classification in machine learning.</div></div>","PeriodicalId":48682,"journal":{"name":"Swarm and Evolutionary Computation","volume":"98 ","pages":"Article 102126"},"PeriodicalIF":8.5,"publicationDate":"2025-08-23","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144890848","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Dual-metric guided multi-strategy hybrid optimization for feature selection on high-dimensional medical data","authors":"Yan Kang ,&nbsp;Dongsheng Zheng ,&nbsp;Haining Wang ,&nbsp;Yue Peng ,&nbsp;Shixuan Zhou","doi":"10.1016/j.swevo.2025.102118","DOIUrl":"10.1016/j.swevo.2025.102118","url":null,"abstract":"<div><div>The high-dimensional feature selection (FS) problem is challenging in medical fields due to the “curse of dimensionality” and the intricate relationships among various features. Although hybrid FS methods achieve high-performance solutions according to various mutual information metrics, such as symmetrical uncertainty (SU) and maximal information coefficient (MIC), they often overlook the differences between these metrics, and are still need to improve search strategies to escape from local optima. To address these challenges, we propose a dual-metric guided multi-strategy hybrid FS method (DGM) for high-dimensional medical datasets. The importance of features are first evaluated based on the SU and MIC metrics, and then the redundancy between features are reduced by fast clustering and grouping strategies. Furthermore, a two-level sampling strategy is proposed to guarantee the diversity and complementarity of population by considering the Jaccard Similarity and the correlation between features. A novel set-based multi-population PSO is designed to collaboratively search the optimal feature subset while obtaining feature importance during the evaluation process by a tri-archive assisted evolution approach. Specifically, two local archives help individuals escape from local optima, while the global archive optimizes the population. Finally, we develop various squeeze-expand mechanisms to dynamically adjust both the search space and the length of individuals to effectively balance exploration and exploitation. The experimental results on 13 medical datasets show that DGM significantly improves classification performance while selecting fewer features. The T-test results further indicate that DGM significantly outperforms all comparison methods in classification performance on 10 datasets, highlighting its strong competitiveness.</div></div>","PeriodicalId":48682,"journal":{"name":"Swarm and Evolutionary Computation","volume":"98 ","pages":"Article 102118"},"PeriodicalIF":8.5,"publicationDate":"2025-08-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144885694","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Hybrid artificial bee colony algorithm with Q-learning for distributed heterogeneous flexible job shop scheduling problem considering machine preventive maintenance","authors":"Rui Wu ,&nbsp;Enzhuang Luo ,&nbsp;Xixing Li ,&nbsp;Hongtao Tang ,&nbsp;Yibing Li","doi":"10.1016/j.swevo.2025.102134","DOIUrl":"10.1016/j.swevo.2025.102134","url":null,"abstract":"<div><div>Current research on preventive maintenance in the scheduling domain predominantly focuses on machine degradation under stable operating conditions. However, the machine works under varying operating conditions (cutting depth, feed rate, etc.) when processing different jobs, and much research ignores the influence of these diverse operating conditions on machine degradation. To address this gap, this paper proposes a novel machine degradation model tailored to various operating conditions and introduces a dual-threshold preventive maintenance strategy, which is integrated with the scheduling problem. To effectively solve this integrated problem, a mixed-integer programming (MIP) framework targeting makespan minimization is constructed, coupled with a hybrid artificial bee colony (ABC) algorithm incorporating a neighborhood search mechanism. First, a three-layer encoding scheme based on factory-machine-operation is designed, and preventive maintenance decisions are incorporated into the decoding strategy. Furthermore, a hybrid population initialization strategy is developed to enhance population diversity. Third, multiple crossover and mutation operators are developed during the employed bee phase, and a simple yet effective operator selection mechanism is employed to improve global search efficiency. In the onlooker bee phase, five neighborhood search operators are proposed to address the local search limitations of traditional ABC algorithms. These operators are adaptively selected via a Q-learning algorithm to strengthen local search performance. Finally, extended computational instances are designed, and comparative experiments validate the effectiveness of the proposed algorithm in solving scheduling problems across different job scales and factory scales.</div></div>","PeriodicalId":48682,"journal":{"name":"Swarm and Evolutionary Computation","volume":"98 ","pages":"Article 102134"},"PeriodicalIF":8.5,"publicationDate":"2025-08-21","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144885695","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Efficient three-stage surrogate-assisted differential evolution for expensive optimization problems","authors":"Abhishek Kumar ,&nbsp;Swagatam Das ,&nbsp;Václav Snášel","doi":"10.1016/j.swevo.2025.102093","DOIUrl":"10.1016/j.swevo.2025.102093","url":null,"abstract":"<div><div>Surrogate-assisted evolutionary algorithms (SAEAs) have received significant acclaim for dealing with intricate and computationally demanding optimization problems. However, a prevalent challenge in many existing algorithms lies in their relatively sluggish convergence in the later stages of optimization. This study introduces an innovative three-stage surrogate-assisted differential evolution (DE) approach that adeptly addresses the demands of early exploration and subsequent exploitation by employing distinct mutation operators and surrogate models. In the initial stage, a combination of radial basis function and multi-dimensional Lipschitz function-assisted DE efficiently identifies a promising region within the complete decision space. After that, the subsequent stage employs a hybrid approach of local and global surrogate-assisted DE, renowned for its robust exploitation capabilities, to hasten the optimization process. This stage incorporates an on-the-fly update approach for both populations and surrogate models, utilizing a pre-set quantity of top-ranked individuals to facilitate updates. Additionally, a sampling technique based on a full-mutation operator is employed to incorporate the best genotypes in the population effectively. A surrogate-assisted local search operator is leveraged in the final stage to optimize the ultimate solution. This stage integrates a radial basis function-based local surrogate function and an interior-point method, enhancing sampling efficiency within the designated local region of interest. The efficacy of the three-stage framework and the proposed strategy is thoroughly validated through simulation experiments, empirical analyses, and ablation studies. Furthermore, we compare the proposed algorithm against other state-of-the-art surrogate-assisted evolutionary algorithms (SAEAs) on a diverse set of expensive benchmark functions and a real-world problem, demonstrating superior performance in terms of both robustness and effectiveness.</div></div>","PeriodicalId":48682,"journal":{"name":"Swarm and Evolutionary Computation","volume":"98 ","pages":"Article 102093"},"PeriodicalIF":8.5,"publicationDate":"2025-08-20","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144865150","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"DGS-EDA: A double-guided sampling estimation of distribution algorithm for multi-robot task assignment as a permutation optimization problem","authors":"Blanca López,&nbsp;Luis Moreno,&nbsp;Concepción A. Monje","doi":"10.1016/j.swevo.2025.102112","DOIUrl":"10.1016/j.swevo.2025.102112","url":null,"abstract":"<div><div>Task assignment refers to the challenge of efficiently allocating tasks, duties, or resources among members of a system. This work investigates the multi-robot task assignment (MRTA) problem, modeled as a variation of the multiple traveling salesman problem (mTSP), and proposes novel approaches based on permutation optimization. An initial study evaluates state-of-the-art evolutionary algorithms (EAs), particularly focusing on estimation of distribution algorithms (EDAs), for their suitability in handling both A-permutation problems, where absolute positioning of the elements within the permutations mostly impact the quality of the solutions, like in the quadratic assignment problem (QAP); and R-permutation problems, where relative positioning dominates, like in the traveling salesman problem (TSP). The adaptation of these algorithms to B-permutation challenges, where both absolute and relative positioning are relevant, such as those presented by the mTSP, has received comparatively limited attention. In this work, addressing this gap led to the creation of a novel double-guided sampling estimation of distribution algorithm (DGS-EDA). The proposed methodologies strategically utilize adjacency relations and consecutive position sampling, guiding the search toward both the least and most observed edges and gene absolute positions to optimize solution paths. Their effectiveness is validated across both problem types; DGS-EDA<span><math><msub><mrow></mrow><mrow><mtext>ew</mtext></mrow></msub></math></span> improves mTSP results by targeting least observed edges, while DGS-EDA<span><math><msub><mrow></mrow><mrow><mtext>eb</mtext></mrow></msub></math></span> enhances TSP outcomes by focusing on the most observed edges. Comprehensive testing using TSPLIB instances demonstrates that the proposed DGS-EDA surpasses existing methods, effectively enhancing the exploration and exploitation of the solution space.</div></div>","PeriodicalId":48682,"journal":{"name":"Swarm and Evolutionary Computation","volume":"98 ","pages":"Article 102112"},"PeriodicalIF":8.5,"publicationDate":"2025-08-19","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144865149","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Adaptive hybrid response mechanism for dynamic multi-objective optimization and its application in multi-robot task allocation","authors":"Tingting Dong ,&nbsp;Peiwen Wang ,&nbsp;Fei Xue ,&nbsp;Yuge Geng ,&nbsp;Zhihua Cui","doi":"10.1016/j.swevo.2025.102123","DOIUrl":"10.1016/j.swevo.2025.102123","url":null,"abstract":"<div><div>The dynamic changes in task requirements and real-time fluctuations in robot states in multi-robot task allocation (MRTA) increase the complexity of algorithm design. This paper presents an Adaptive Multi-Objective Evolutionary Algorithm with Hybrid Response Mechanism (AMOEAD-HRM) for dynamic multi-objective MRTA, addressing environmental uncertainty through innovative mechanisms. AMOEAD-HRM proposes a GNG-based prediction response mechanism, leveraging Growing Neural Gas (GNG) networks to model the time-varying nature of tasks and robot states. Unlike fixed-architecture predictors, GNG captures data topological structures to construct adaptive predictive models, dynamically adjusting to fluctuations and uncertainties by iteratively optimizing network topology. This enables effective characterization of complex temporal patterns without prior distribution assumptions, providing a robust foundation for predicting dynamic changes. To enhance responsiveness, the algorithm integrates a memory-based response mechanism and a Gaussian polynomial mixture mutation strategy. A dynamic adaptive weight adjustment strategy selects optimal response mechanisms according to environmental variation degrees, balancing prediction accuracy and real-time adaptability to improve system robustness and flexibility. Experimental validation on 19 benchmark problems shows AMOEAD-HRM’s superiority. In dynamic scenarios, it responds 46.1% faster than DNSGA-II. Under high dynamics, its solution sets have 3.4% higher MHV than DNSGA-II. With moderate changes, MHV is 0.34% higher than SGEA.</div></div>","PeriodicalId":48682,"journal":{"name":"Swarm and Evolutionary Computation","volume":"98 ","pages":"Article 102123"},"PeriodicalIF":8.5,"publicationDate":"2025-08-16","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144858250","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Parallel genetic algorithms on hybrid servers: Design, implementation, and optimization for performance and energy","authors":"Amr Abdelhafez ,&nbsp;Ravi Reddy Manumachu ,&nbsp;Alexey Lastovetsky","doi":"10.1016/j.swevo.2025.102110","DOIUrl":"10.1016/j.swevo.2025.102110","url":null,"abstract":"<div><div>Parallel Genetic Algorithms (PGAs) have been widely applied to accelerate solutions for real-world problems such as energy optimization in building constructions, data preprocessing and model selection steps in data mining, real-time control of multilevel inverters in electronics, land-use planning, nanoscience, optimal power flow in power systems, and road traffic management.</div><div>The state-of-the-art research proposes PGAs optimized solely for performance and for solving optimization problems on a multicore CPU, GPU, or clusters of multicore CPUs. However, no research has analyzed PGAs for heterogeneous hybrid platforms comprising multicore CPUs and multiple accelerators that utilize all computing devices in parallel. Furthermore, no definitive comparative research comprehensively investigates the energy consumption of PGAs in hybrid systems versus multicore CPUs or GPUs.</div><div>We address the above gaps in the prior art in this work. First, we present a novel parallelization approach (HPIGA) tailored for heterogeneous hybrid platforms, featuring a portable implementation that utilizes all available computational devices, including multicore CPUs and GPUs. We conduct a comprehensive investigation into the performance and energy profiles of this approach. We compare it with three other traditional parallel approaches across a range of dimensions, varying from 100 dimensions and up to 5000 dimensions. The results showed HPIGA’s competitive energy consumption behavior and promising performance compared to other traditional approaches under the study.</div><div>Moreover, we formulate a bi-objective optimization problem of a PGA employing a parallel island model and executing on a hybrid server comprising <span><math><mi>p</mi></math></span> compute devices. The problem has two objectives: performance and energy. The decision variable used in our bi-objective optimization problem is workload distribution, which is proportional to the number of islands. We study the efficacy of our proposed PGA on a hybrid server platform with an Intel Icelake multicore CPU and two Nvidia A40 GPUs, analyzing execution time and dynamic energy profiles under two power governors. The resulting Pareto front graphs provide valuable insights, serving as crucial benchmarks for the future development and use of efficient, energy-aware optimization techniques across diverse computational devices.</div></div>","PeriodicalId":48682,"journal":{"name":"Swarm and Evolutionary Computation","volume":"98 ","pages":"Article 102110"},"PeriodicalIF":8.5,"publicationDate":"2025-08-16","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144858324","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"A robust counterpart model and matheuristic-oriented evolutionary algorithm for evaluating energy consumption of project scheduling with uncertainty","authors":"Zheng Gao ,&nbsp;Liping Zhang ,&nbsp;Zikai Zhang ,&nbsp;Zixiang Li ,&nbsp;Yingli Li","doi":"10.1016/j.swevo.2025.102124","DOIUrl":"10.1016/j.swevo.2025.102124","url":null,"abstract":"<div><div>Makespan is a key metric for evaluating project progress, while energy consumption directly impacts green performance metrics. These are the key metrics that managers focus on. Based on this, an energy-aware multi-mode resource-constrained project scheduling problem is proposed. However, activity durations in real project scheduling are often uncertain. Energy consumption and makespan cannot be accurately evaluated due to uncertain activity durations. In response to this, a multi-objective mixed-integer linear programming (MILP) model is proposed to trade off makespan and total energy consumption with uncertainty. Then, the uncertainty level and reliability level are introduced to quantify uncertain activity durations. Finally, the MILP model is transformed into a robust counterpart model to obtain robust non-dominated solutions for small-scale instances. Additionally, a matheuristic-oriented multi-objective evolutionary algorithm is designed to address large-scale instances. Finally, extensive numerical experiments are conducted to validate the proposed model and algorithm. The experimental results demonstrate that the robust counterpart model can quickly obtain a set of robust non-dominated solutions for small-scale instances. The matheuristic local optimization approach can indeed rapidly improve the quality of robust non-dominated solutions. Furthermore, the matheuristic-oriented multi-objective evolutionary algorithm outperforms state-of-the-art algorithms in terms of several multi-objective evaluation indicators.</div></div>","PeriodicalId":48682,"journal":{"name":"Swarm and Evolutionary Computation","volume":"98 ","pages":"Article 102124"},"PeriodicalIF":8.5,"publicationDate":"2025-08-16","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144858251","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"A constrained multi-objective evolutionary algorithm for multi-class instance selection","authors":"Qijun Wang,&nbsp;Yujie Ge,&nbsp;Lei Zhang,&nbsp;Fan Cheng","doi":"10.1016/j.swevo.2025.102120","DOIUrl":"10.1016/j.swevo.2025.102120","url":null,"abstract":"<div><div>As a data processing technology, instance selection (IS) aims to select a small number of instances with the same (or even higher) classification capability. Due to its widely applications, many IS algorithms with promising performance have been suggested. Despite that, most of existing algorithms focus on designing new IS algorithms by using different optimizing techniques, and few of them consider the imbalance among different classes in multi-class IS. To address the problem, in this paper, a constrained optimization problem is firstly formulated for multi-class IS, where the “hard constraint” and the “soft constraint” are defined to model the multi-class IS problem more accurately. Then, to solve the constrained optimization problem, a multi-objective evolutionary algorithm termed as CMOEA-MIS is proposed, by which the instance subsets with high quality could be achieved. Specifically, in CMOEA-MIS, a constraint-based solution selection strategy is developed based on the dominance relationship that considers both constraint violation and the quality of solution, and is introduced to choose the individuals in the mating pool. In addition, a two-stage based mutation strategy is also suggested in CMOEA-MIS, by which the quality of the final obtained instance subsets is further improved. Experimental results on the multi-class datasets with different characteristics have demonstrated that CMOEA-MIS can obtain multi-class instance subsets with more than 50% reduction rate, and can ensure the accuracy of each class, and can be used to train the classifiers with comparable or better performance than the state-of-the-art IS algorithms.</div></div>","PeriodicalId":48682,"journal":{"name":"Swarm and Evolutionary Computation","volume":"98 ","pages":"Article 102120"},"PeriodicalIF":8.5,"publicationDate":"2025-08-15","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144852944","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}