Swarm and Evolutionary Computation最新文献_第9页

A constrained-neighborhood memetic algorithm with diversity control for the JSP-TWT 基于分集控制的JSP-TWT约束邻域模因算法

IF 8.5 1区计算机科学

Swarm and Evolutionary Computation Pub Date : 2026-06-01 Epub Date: 2026-06-05 DOI: 10.1016/j.swevo.2026.102429

Zhiyong Ba , Jinduo Liu , Yiping Yuan

{"title":"A constrained-neighborhood memetic algorithm with diversity control for the JSP-TWT","authors":"Zhiyong Ba , Jinduo Liu , Yiping Yuan","doi":"10.1016/j.swevo.2026.102429","DOIUrl":"10.1016/j.swevo.2026.102429","url":null,"abstract":"<div><div>The job shop scheduling problem with the objective of minimizing total weighted tardiness (JSP-TWT) is structurally challenging, since tight due dates or heterogeneous job weights often induce multiple critical chains and make the objective highly sensitive to local perturbations. To address this issue, this paper proposes a constrained-neighborhood memetic algorithm for JSP-TWT, denoted by MA-CN. Under an extended disjunctive-graph framework, a critical-tree representation is introduced to characterize tardiness-related critical structures, based on which a problem-specific constrained neighborhood, CN7-TWT, is developed. Candidate moves are filtered before evaluation by unified feasibility conditions and structural elimination rules for non-improving moves, thereby reducing neighborhood redundancy. Furthermore, CN7-TWT is embedded into tabu search as the core neighborhood operator for local intensification, while a dynamic score-driven population updating strategy is used to preserve diversity and alleviate premature convergence at the population level. Computational results on the LA01–LA40 benchmark set show that the proposed method substantially reduces neighborhood search cost while maintaining competitive local-search quality, and delivers the best overall performance among the compared algorithms. In particular, MA-CN updates 25 best-known solutions under different due-date tightness settings, demonstrating its effectiveness and robustness for JSP-TWT.</div></div>","PeriodicalId":48682,"journal":{"name":"Swarm and Evolutionary Computation","volume":"106 ","pages":"Article 102429"},"PeriodicalIF":8.5,"publicationDate":"2026-06-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148179792","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}

引用次数: 0

Fairness-constrained influence maximisation via multi-objective optimisation 通过多目标优化实现公平约束下的影响力最大化

IF 8.5 1区计算机科学

Swarm and Evolutionary Computation Pub Date : 2026-06-01 Epub Date: 2026-06-02 DOI: 10.1016/j.swevo.2026.102428

Ziying Zhao , Weihua Li , Jing Ma , Jianhua Jiang , Quan Bai , Wen Gu

{"title":"Fairness-constrained influence maximisation via multi-objective optimisation","authors":"Ziying Zhao , Weihua Li , Jing Ma , Jianhua Jiang , Quan Bai , Wen Gu","doi":"10.1016/j.swevo.2026.102428","DOIUrl":"10.1016/j.swevo.2026.102428","url":null,"abstract":"<div><div>The Influence Maximisation (IM) problem seeks to select a set of seed nodes to maximise information diffusion in a network. While existing approaches have achieved significant improvements in overall diffusion, they often overlook fairness across communities, which can result in biased dissemination and the exclusion of disadvantaged groups. To address this, we define the Fair Multi-objective Influence Maximisation (FMOIM) problem, which jointly optimises influence spread and equity fairness. Equity Fairness is modelled at the community level as the alignment between the realised diffusion-benefit distribution and a desired reference allocation. Jensen–Shannon divergence (JSD) similarity quantifies distributional deviation from the reference allocation, while Jain’s fairness index characterises the evenness of benefit allocation across communities. To solve FMOIM, FairWolf is proposed as a problem-driven discrete multi-objective optimisation model for fairness-aware influence maximisation. It reformulates the Grey Wolf Optimiser dynamics to search directly over fixed-budget seed sets under community-level fairness objectives, capturing the spread and fairness trade-off. FairWolf incorporates three components: (i) a discrete position-updating mechanism tailored to seed-set construction, (ii) an Explorer-Augmented Leader Selection strategy that enhances population diversity while maintaining convergence pressure, and (iii) a Hypervolume (HV)-triggered perturbation mechanism that adaptively mitigates stagnation in non-convex multi-objective search spaces. Experiments on eight real-world networks demonstrate that the FairWolf model consistently outperforms state-of-the-art baselines, yielding a higher HV value and more uniformly distributed Pareto fronts. These results demonstrate its effectiveness and practicality for fairness-aware diffusion in applications such as viral marketing, public health, and resource allocation.</div></div>","PeriodicalId":48682,"journal":{"name":"Swarm and Evolutionary Computation","volume":"106 ","pages":"Article 102428"},"PeriodicalIF":8.5,"publicationDate":"2026-06-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148179793","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}

引用次数: 0

Feedback-driven adaptive variable grouping for decomposition-based large-scale multiobjective optimization 基于分解的大规模多目标优化的反馈驱动自适应变量分组

IF 8.5 1区计算机科学

Swarm and Evolutionary Computation Pub Date : 2026-06-01 Epub Date: 2026-05-30 DOI: 10.1016/j.swevo.2026.102416

Qi Liu , Rui Lin , Li Lyu

{"title":"Feedback-driven adaptive variable grouping for decomposition-based large-scale multiobjective optimization","authors":"Qi Liu , Rui Lin , Li Lyu","doi":"10.1016/j.swevo.2026.102416","DOIUrl":"10.1016/j.swevo.2026.102416","url":null,"abstract":"<div><div>Large-scale multiobjective optimization problems involving hundreds or thousands of decision variables pose severe challenges to conventional multiobjective evolutionary algorithms due to the exponential growth of the search space. Decision variable analysis has emerged as a promising strategy for addressing such problems by decomposing decision variables into convergence-related and diversity-related groups. However, existing techniques suffer from two critical shortcomings: the high computational cost of interaction detection and the static nature of grouping throughout evolution, which fails to adapt to the changing landscape characteristics during optimization. Although dynamic regrouping and adaptive operator selection have been explored separately in prior work, no existing method integrates performance-driven regrouping with group-level operator adaptation within a unified decomposition-based framework This paper proposes MOEA/D-FAVG, a feedback-driven adaptive variable grouping algorithm embedded in a decomposition-based multiobjective evolutionary framework. Three key mechanisms are introduced: a lightweight gradient-approximation-based variable classification that requires significantly fewer function evaluations than perturbation-based approaches; a feedback-driven regrouping mechanism that dynamically adjusts the grouping structure based on subgroup-level performance indicators; and a group-level adaptive operator selection strategy that assigns exploitation- or exploration-oriented operators to each subgroup according to its current optimization state. Comprehensive experiments on the LSMOP and DTLZ benchmark suites with 200 to 5000 decision variables demonstrate that MOEA/D-FAVG achieves statistically superior performance over nine state-of-the-art large-scale multiobjective evolutionary algorithms in terms of both inverted generational distance and hypervolume.</div></div>","PeriodicalId":48682,"journal":{"name":"Swarm and Evolutionary Computation","volume":"106 ","pages":"Article 102416"},"PeriodicalIF":8.5,"publicationDate":"2026-06-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148179794","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}

引用次数: 0

Meta-Black-Box Optimization with ensemble surrogate modeling for robustness–accuracy trade-off within SAEA 基于集成代理模型的元黑盒优化，在SAEA中实现鲁棒性与准确性的权衡

IF 8.5 1区计算机科学

Swarm and Evolutionary Computation Pub Date : 2026-06-01 Epub Date: 2026-06-10 DOI: 10.1016/j.swevo.2026.102425

Xiao Jin , Yongxiong Wang , Haobo Liu , Yudong Du , Yukun Du

{"title":"Meta-Black-Box Optimization with ensemble surrogate modeling for robustness–accuracy trade-off within SAEA","authors":"Xiao Jin , Yongxiong Wang , Haobo Liu , Yudong Du , Yukun Du","doi":"10.1016/j.swevo.2026.102425","DOIUrl":"10.1016/j.swevo.2026.102425","url":null,"abstract":"<div><div>Surrogate-Assisted Evolutionary Algorithms (SAEAs) have been widely used for expensive black-box optimization problems. However, their reliance on rigid and manually designed components limits flexibility and generalization across tasks. Meta-Black-Box Optimization (MetaBBO) provides a promising solution by adaptively configuring algorithmic components. Yet, existing MetaBBO approaches typically control only a single component, and few studies address the unified control of multi-component optimizers like SAEA. Moreover, current research rarely considers the robustness–accuracy trade-off in surrogate modeling, which is essential for stable early exploration and accurate late-stage exploitation. To address these issues, we propose AdaE-SAEA (Adaptive Ensemble Surrogate-Assisted Evolutionary Algorithm) for expensive multi-objective optimization. AdaE-SAEA integrates SAEA as the low-level optimizer within the MetaBBO framework and jointly controls both the infill criterion and ensemble-based surrogate modeling. Bagging and boosting are designed as surrogate modeling modules to adaptively balance robustness and accuracy during different search phases, while the meta-policy simultaneously controls the infill criterion to ensure adaptive sampling decisions. The meta-policy is trained via reinforcement learning with parallel sampling and centralized training for improved efficiency and transferability. Experiments on synthetic and real-world problems show that AdaE-SAEA outperforms state-of-the-art baselines and MetaBBO-based methods. Additionally, we demonstrate the effectiveness of TabPFN as the base surrogate model for ensemble learning. To the best of our knowledge, this is the first work to unify the control of surrogate modeling and infill criteria in SAEA while explicitly addressing the robustness–accuracy trade-off.</div></div>","PeriodicalId":48682,"journal":{"name":"Swarm and Evolutionary Computation","volume":"106 ","pages":"Article 102425"},"PeriodicalIF":8.5,"publicationDate":"2026-06-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148242412","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}

引用次数: 0

Data-driven whale optimization for high-dimensional expensive problems using hybrid local search strategy 基于混合局部搜索策略的高维昂贵问题数据驱动鲸鱼优化

IF 8.5 1区计算机科学

Swarm and Evolutionary Computation Pub Date : 2026-06-01 Epub Date: 2026-06-02 DOI: 10.1016/j.swevo.2026.102427

Huachao Dong, Xiao-Yao Han

{"title":"Data-driven whale optimization for high-dimensional expensive problems using hybrid local search strategy","authors":"Huachao Dong, Xiao-Yao Han","doi":"10.1016/j.swevo.2026.102427","DOIUrl":"10.1016/j.swevo.2026.102427","url":null,"abstract":"<div><div>High-dimensional expensive optimization problems present inherent challenges due to the computational cost of function evaluations and the curse of dimensionality. In this paper, we propose a novel Data-driven Whale Optimization Algorithm with a Hybrid Local Search strategy (DWOA-HLS). The algorithm features a cyclic search structure that systematically coordinates a global metaheuristic search phase with a hybrid local search phase, all guided by radial basis function surrogate models. The proposed approach incorporates a hybrid local strategy that alternately employs a local metaheuristic search for robust regional exploration and an SQP-based gradient refinement for accelerated convergence. Comprehensive experiments on benchmarks demonstrate that DWOA-HLS significantly outperforms several surrogate-free and surrogate-assisted evolutionary algorithms, showing particularly superior performance as dimensionality increases. An ablation study confirms the critical contribution of the hybrid local search, and further analysis reveals the high computational efficiency of DWOA-HLS. The effectiveness of DWOA-HLS is further validated through a real-world case involving the hydrodynamic shape optimization of a blended-wing-body underwater glider.</div></div>","PeriodicalId":48682,"journal":{"name":"Swarm and Evolutionary Computation","volume":"106 ","pages":"Article 102427"},"PeriodicalIF":8.5,"publicationDate":"2026-06-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148179795","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}

引用次数: 0

A constrained multi-objective co-evolutionary algorithm with a dynamic diversity-enhanced auxiliary population 一种具有动态多样性增强辅助种群的约束多目标协同进化算法

IF 8.5 1区计算机科学

Swarm and Evolutionary Computation Pub Date : 2026-06-01 Epub Date: 2026-06-02 DOI: 10.1016/j.swevo.2026.102426

Zhiqiang Zeng , Ye Tian , Xiaohui Yan , Liangwei Zhang , Jianyu Long , Qiang Luo , Chuangquan Chen

{"title":"A constrained multi-objective co-evolutionary algorithm with a dynamic diversity-enhanced auxiliary population","authors":"Zhiqiang Zeng , Ye Tian , Xiaohui Yan , Liangwei Zhang , Jianyu Long , Qiang Luo , Chuangquan Chen","doi":"10.1016/j.swevo.2026.102426","DOIUrl":"10.1016/j.swevo.2026.102426","url":null,"abstract":"<div><div>Balancing diversity, convergence, and feasibility has long been a challenge in constrained multi-objective evolutionary algorithms. To address this issue, this study proposes a constrained multi-objective co-evolutionary algorithm with a dynamic diversity-enhanced auxiliary population (CMOCEA-DDAP). First, we construct a main population that simultaneously considers diversity, convergence, and feasibility, along with an auxiliary population focused on diversity and convergence. Second, we develop a co-evolutionary method that employs a diversity sampling strategy to sample non-dominated solutions from the main population; the sampled solutions are then used to guide the evolution of the auxiliary population, which in turn enhances the diversity of the main population. Third, we propose an adaptive multi-strategy search algorithm based on dual ranking with differences in preference between constraints and objectives. Fourth, we introduce a cooperative environmental selection strategy to balance objectives and constraints. The performance of CMOCEA-DDAP is evaluated on 56 benchmark test problems and 21 real-world problems. Experimental results show that, compared with 14 state-of-the-art constrained multi-objective algorithms, CMOCEA-DDAP achieves significantly better performance on multiple metrics across both benchmark and real-world problems.</div></div>","PeriodicalId":48682,"journal":{"name":"Swarm and Evolutionary Computation","volume":"106 ","pages":"Article 102426"},"PeriodicalIF":8.5,"publicationDate":"2026-06-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148179791","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}

引用次数: 0

Artificial bee colony meets variable neighborhood search: Fresh insights into influence maximization in social networks 人工蜂群满足可变邻域搜索：对社会网络影响最大化的新见解

IF 8.5 1区计算机科学

Swarm and Evolutionary Computation Pub Date : 2026-05-01 Epub Date: 2026-05-05 DOI: 10.1016/j.swevo.2026.102404

Jianxin Tang , Fangjie Dong , Chen Wu , Wenhuan Chen , Huiyi Wei

{"title":"Artificial bee colony meets variable neighborhood search: Fresh insights into influence maximization in social networks","authors":"Jianxin Tang , Fangjie Dong , Chen Wu , Wenhuan Chen , Huiyi Wei","doi":"10.1016/j.swevo.2026.102404","DOIUrl":"10.1016/j.swevo.2026.102404","url":null,"abstract":"<div><div>As one of the fundamental topics within complex networks, the influence maximization problem invites the pursuit of key nodes that can maximize the influence spread. However, conventional meta-heuristics struggle to identify an optimal seed set due to rigid neighborhood generation strategies and weak directional guidance. In light of this, the artificial bee colony algorithm empowered by the variable neighborhood search strategy is deployed to address the above challenges. Specifically, an elite-global cooperative neighborhood construction mechanism that injects directional cues is devised to avoid aimless exploration. Then, a variable neighborhood search strategy that dynamically switches among a continuum of fine-grained neighborhood structures is introduced to achieve a good trade-off between exploration and exploitation. Furthermore, when the population converges prematurely to local optima, a fine-grained, individual-centric stagnation control mechanism combined with a stagnation-aware population management mechanism further maintains search vitality at both the individual and population levels. Experiments on nine real-world networks reveal that the proposed algorithm achieves an overall average improvement of 6.3% in influence spread relative to all baseline methods, demonstrating its strong scalability across different types of networks. Such findings showcase its potential for practical applications in rumor control and cascading failures.</div></div>","PeriodicalId":48682,"journal":{"name":"Swarm and Evolutionary Computation","volume":"105 ","pages":"Article 102404"},"PeriodicalIF":8.5,"publicationDate":"2026-05-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"147860012","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}

引用次数: 0

An evolutionary game-based optimization approach for multi-tier collaborative capacity allocation in industrial chains with chain master enterprise 基于演化博弈的产业链主企业多层协同产能配置优化方法

IF 8.5 1区计算机科学

Swarm and Evolutionary Computation Pub Date : 2026-05-01 Epub Date: 2026-05-08 DOI: 10.1016/j.swevo.2026.102408

Mingzhe She , Xiao Lai , Yanshan Gao , Chenze Wang , Xiaohan Zhang , Zhifang Xu , Min Liu , Fei Qiao

{"title":"An evolutionary game-based optimization approach for multi-tier collaborative capacity allocation in industrial chains with chain master enterprise","authors":"Mingzhe She , Xiao Lai , Yanshan Gao , Chenze Wang , Xiaohan Zhang , Zhifang Xu , Min Liu , Fei Qiao","doi":"10.1016/j.swevo.2026.102408","DOIUrl":"10.1016/j.swevo.2026.102408","url":null,"abstract":"<div><div>Faced with escalating political and economic complexities and intense market competition, efficient allocation of capacity across tiers has become a strategic imperative for the chain master enterprise seeking to maximize system-wide resource utilization. However, complex interdependencies between upstream and downstream firms, combined with the strategic preferences of the chain master enterprise, have posed significant challenges for the performance of most strategies for single-tier capacity planning. This highlights the necessity for developing effective collaborative approaches to multi-tier capacity allocation that address performance optimization enhancement from a system-wide perspective. To address the above issues, this paper proposes a novel collaborative capacity allocation approach for the industrial chain under chain master enterprise-led governance. A general three-tier collaboration model is established by integrating an explicit network representation of cross-tier constraints and interactions, and a collaborative capacity allocation problem via the collaboration network model is formulated. To incorporate the chain master enterprise’s strategic intent into resource allocation, preference coefficients are introduced to guide the load distribution pattern. Building upon this, an evolutionary game-based collaborative capacity allocation approach is developed, which iteratively integrates the network structural evolution and parameter evolution to make capacity allocation decisions, thereby significantly improving cross-tier collaboration effectiveness and optimizing the operational benefits of the industrial chain. Comparative experiments validate the superiority of this evolutionary game-based approach in these areas, confirming its effectiveness in optimizing capacity allocation in the industrial chain.</div></div>","PeriodicalId":48682,"journal":{"name":"Swarm and Evolutionary Computation","volume":"105 ","pages":"Article 102408"},"PeriodicalIF":8.5,"publicationDate":"2026-05-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"147860010","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}

引用次数: 0

Data-driven estimation of printed circuit heat exchanger internal geometric data via hybrid genetic algorithm and machine learning method 基于混合遗传算法和机器学习方法的印制电路换热器内部几何数据的数据驱动估计

IF 8.5 1区计算机科学

Swarm and Evolutionary Computation Pub Date : 2026-05-01 Epub Date: 2026-05-05 DOI: 10.1016/j.swevo.2026.102400

Feride Kutbay, Senem Şentürk Lüle

{"title":"Data-driven estimation of printed circuit heat exchanger internal geometric data via hybrid genetic algorithm and machine learning method","authors":"Feride Kutbay, Senem Şentürk Lüle","doi":"10.1016/j.swevo.2026.102400","DOIUrl":"10.1016/j.swevo.2026.102400","url":null,"abstract":"<div><div>Printed Circuit Heat Exchangers (PCHEs) are used in advanced thermal systems, including supercritical CO₂ (s-CO₂) cycles for nuclear reactors, waste heat recovery units, and concentrated solar power (CSP) plants due to their superiority on compactness and effectiveness. However, the internal geometric configurations of commercially available PCHEs are often confidential, limiting accurate modeling, simulation, and control model development in experimental test rigs and operational facilities. To address this issue, this study proposes a reverse engineering methodology to estimate the unknown internal configuration of an existing PCHE from experimental data by using a penalty function-based genetic algorithm (GA) coupled with K-Means clustering. An indigenous steady-state discretized logarithmic mean temperature difference (LMTD) heat exchanger code NODEX which can calculate the performance of a variable heat sink arrangement of s-CO<sub>2</sub> heat exchanger for any given set of boundary conditions was used in GA simulations. Unlike conventional optimization studies aiming efficiency, size, or cost, the proposed methodology minimizes the discrepancy between experimentally measured and numerically calculated data by modifying the geometric data. The validated GA analysis resulted in 102 possible PCHE configurations. The K-Means clustering grouped theses into 5 clusters. The cluster with the highest number of configurations and the lowest penalty function was selected as the final configuration. The results showed that the discrepancies in key variables such as channel diameter and the number of hot and cold channels were found to be <1%, demonstrating a high level of accuracy of the proposed methodology.</div></div>","PeriodicalId":48682,"journal":{"name":"Swarm and Evolutionary Computation","volume":"105 ","pages":"Article 102400"},"PeriodicalIF":8.5,"publicationDate":"2026-05-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"147860011","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}

引用次数: 0

Enhanced multiclass cardiovascular disease detection via Kernel-PSO feature selection: a comparative benchmark of gradient-driven and bio-inspired machine learning algorithms 通过核粒子群特征选择增强多类心血管疾病检测：梯度驱动和生物启发机器学习算法的比较基准

IF 8.5 1区计算机科学

Swarm and Evolutionary Computation Pub Date : 2026-05-01 Epub Date: 2026-05-09 DOI: 10.1016/j.swevo.2026.102405

Onder Tutsoy , Hilmi Erdem Sumbul

{"title":"Enhanced multiclass cardiovascular disease detection via Kernel-PSO feature selection: a comparative benchmark of gradient-driven and bio-inspired machine learning algorithms","authors":"Onder Tutsoy , Hilmi Erdem Sumbul","doi":"10.1016/j.swevo.2026.102405","DOIUrl":"10.1016/j.swevo.2026.102405","url":null,"abstract":"<div><div>The global burden of the cardiovascular disease is compounded by a critical diagnostic gap, where the traditional electrocardiogram (ECG) interpretation often fails to resolve overlapping multiclass feature spaces. This paper addresses these inconsistencies by proposing feature selection-based robust machine learning algorithms for the automated detection of the Arrhythmia (ARR), Congestive Heart Failure (CHF), Atrial Fibrillation (AF), and Normal Sinus Rhythm (NSR). By introducing a performance-driven Kernel-Particle Swarm Optimization (Kernel-PSO) approach with a polynomial basis function, the research replaces the traditional accuracy-centric metrics with an objective function that eliminates the empirical trial-and-error process and provides a precise mathematical basis for the optimal features to detect multiclass diseases. Experimental results demonstrate that the objective function decreases monotonically as the feature count increases, consistently identifying the most informative features across various subsets. A rigorous benchmarking analysis reveals that the proposed feature selection-based Batch Least Squares (BLS) and Non-negative Least Squares (NNLS) significantly outperform the bio-inspired Firefly machine learning algorithm in terms of convergence stability and testing robustness. Specifically, the BLS provides quantified proof of superior performance when evaluated with the unseen data, achieving a F1-score of 0.98, a significantly low RMSE of 0.04, and a superior Cohen’s Kappa value of 0.96, thereby ensuring high diagnostic reliability and generalization capability across the multiclass dataset.</div></div>","PeriodicalId":48682,"journal":{"name":"Swarm and Evolutionary Computation","volume":"105 ","pages":"Article 102405"},"PeriodicalIF":8.5,"publicationDate":"2026-05-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"147860065","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}

引用次数: 0