Swarm and Evolutionary Computation最新文献

{"title":"An evolutionary method with shift pattern learning for real-world multi-skilled personnel scheduling with flexible shifts","authors":"Ning Xue,&nbsp;Ruibin Bai,&nbsp;Huan Jin,&nbsp;Tianxiang Cui","doi":"10.1016/j.swevo.2025.102160","DOIUrl":"10.1016/j.swevo.2025.102160","url":null,"abstract":"<div><div>Personnel scheduling remains a significant organizational challenge with substantial potential for cost and time savings. Despite extensive research in this domain, few studies have been successfully implemented in practice, and even fewer have gained widespread acceptance among end-users. This gap between research and application often arises from oversimplified real-world models, which may result from subjective solution evaluations or a lack of collaboration between modelers and end-users. To bridge this gap, this paper proposes a machine learning-enhanced memetic algorithm (MLMA) that mimics schedules created by experts to solve a highly complex personnel scheduling problem involving multi-skilled workers and flexible shift types (irregular workforce)—a real-world challenge commonly faced in the hospitality sector. By leveraging historical scheduling preferences, the MLMA generates solutions that align with past practices, enhancing their practicality and appeal to end-users. Experiments conducted on real-life instances demonstrate the effectiveness of the proposed approach in addressing real-world problems, where the workforce is predominantly part-time, possesses mixed skills, and requires flexible shifts. Furthermore, the results highlight the MLMA’s ability to identify shift patterns that closely resemble historical schedules, underscoring its potential for practical implementation and its role in bridging the gap between research and real-world application.</div></div>","PeriodicalId":48682,"journal":{"name":"Swarm and Evolutionary Computation","volume":"99 ","pages":"Article 102160"},"PeriodicalIF":8.5,"publicationDate":"2025-09-13","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145049197","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 metaheuristic algorithms for image watermarking: An experimental study","authors":"Anna Melman,&nbsp;Oleg Evsutin","doi":"10.1016/j.swevo.2025.102163","DOIUrl":"10.1016/j.swevo.2025.102163","url":null,"abstract":"<div><div>Invisible image watermarking is a promising method for protecting the copyright of digital images such as photographs, illustrations, and scans. An effective watermarking algorithm embeds a special mark into an image that does not change the image content but can be extracted from it even after some common post-processing operations such as cropping or compression. Many authors use metaheuristic optimization algorithms to achieve a trade-off between imperceptibility and robustness of embedding. In recent years, researchers have been interested in hybrid metaheuristics, which combine operations of individual metaheuristics in some way. However, designs and compositions of hybrid metaheuristic optimization schemes for image watermarking have not been sufficiently studied to date. In this paper, we present an experimental study of various hybrid metaheuristics including sequential, interleaved, and parallel schemes for popular bioinspired optimization algorithms including genetic algorithm, differential evolution algorithm, particle swarm optimization algorithm, firefly algorithm, and artificial bee colony algorithm. We evaluate the effectiveness of hybrid metaheuristics for image watermarking using an algorithm based on changing the ratio between absolute values ​​of sums of discrete cosine transform coefficient groups as an example and perform an experimental comparison of different schemes. The results of the study show that a approach to metaheuristic hybridization and a composition of hybrid scheme significantly affect the imperceptibility and robustness of the image watermarking algorithm. In particular, the interleaved hybridization type provides the best results for the algorithm under consideration.</div></div>","PeriodicalId":48682,"journal":{"name":"Swarm and Evolutionary Computation","volume":"99 ","pages":"Article 102163"},"PeriodicalIF":8.5,"publicationDate":"2025-09-13","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145049772","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 learning-and-knowledge-assisted multi-population collaborative evolutionary algorithm for integrated design-production-distribution scheduling problems in mass personalized customization","authors":"Yanhe Jia ,&nbsp;Wei Wang ,&nbsp;Jian Zhang","doi":"10.1016/j.swevo.2025.102158","DOIUrl":"10.1016/j.swevo.2025.102158","url":null,"abstract":"<div><div>Recently, new requirements are proposed for the manufacturing industry transitioning to distributed production models due to emergence of mass personalized customization. Integrated scheduling of design, production and distribution, mixed management of batch and flexible manufacturing are becoming the imminent challenges faced by enterprises. This article proposes an integrated design-production-distribution scheduling problem in distributed mixed shops. It considers distributed flow shops for batch manufacturing and distributed flexible job shops for flexible manufacturing. First, a mixed integer linear programming model is formulized to minimize the maximum completion time, total costs, and total tardiness. Second, a learning-and-knowledge-assisted multi-population collaborative evolutionary algorithm is developed to settle the model. Genetic operators are adopted to improve the global and local search abilities. Three subpopulations with adaptive crossover and mutation probabilities are constructed to enhance the convergence and diversity of population. A Q-learning-assisted cooperative approach is adopted to realize the information communication among subpopulations in the genetic operations. The Q-learning method is used to intelligently choose parent individuals from three subpopulations by utilizing its self-learning strategies. A variable neighborhood search approach considering problem-knowledge neighborhood structures is devised to refine the excellent individuals in population. Finally, the presented algorithm is compared against three well-known intelligent optimization methods on a collection of instances. Comparison outcomes verify the superiority of the developed algorithm in handling the considered problem.</div></div>","PeriodicalId":48682,"journal":{"name":"Swarm and Evolutionary Computation","volume":"99 ","pages":"Article 102158"},"PeriodicalIF":8.5,"publicationDate":"2025-09-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145049200","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":"Using genetic programming to improve data collection for offline reinforcement learning","authors":"David Halder,&nbsp;Georgios Douzas,&nbsp;Fernando Bacao","doi":"10.1016/j.swevo.2025.102140","DOIUrl":"10.1016/j.swevo.2025.102140","url":null,"abstract":"<div><div>Offline Reinforcement Learning (RL) learns policies solely from fixed pre-collected datasets, making it applicable to use-cases where data collection is expensive or risky. Consequently, the performance of these offline learners is highly dependent on the dataset used. Still the questions of how this data is collected and what dataset characteristics are needed are not thoroughly investigated. Simultaneously, evolutionary methods have reemerged as a promising alternative to classic RL, leading to the field of evolutionary RL (EvoRL), combining the two learning paradigms to exploit their supplementary attributes. This study aims to join these research directions and examine the effects of Genetic Programming (GP) on dataset characteristics in RL and its potential to enhance the performance of offline RL algorithms. A comparative approach was employed, comparing Deep Q-Networks (DQN) and GP for data collection across multiple environments and collection modes. The exploration and exploitation capabilities of these methods were quantified and a comparative analysis was conducted to determine whether data collected through GP led to superior performance in multiple offline learners. The findings indicate that GP demonstrates strong and stable performance in generating high-quality experiences with competitive exploration. GP exhibited lower uncertainty in experience generation compared to DQN and produced high trajectory quality datasets across all environments. More offline algorithms showed statistically significant performance gains with GP-collected data than trained on DQN-collected trajectories. Furthermore, their performance was less dependent on the environment, as the GP consistently generated high-quality datasets. This study showcases the effective combination of GP's properties with offline learners, suggesting a promising avenue for future research in optimizing data collection for RL.</div></div>","PeriodicalId":48682,"journal":{"name":"Swarm and Evolutionary Computation","volume":"99 ","pages":"Article 102140"},"PeriodicalIF":8.5,"publicationDate":"2025-09-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145049199","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":"An effective combination of mechanisms for particle swarm optimization-based ensemble strategy","authors":"Libin Hong ,&nbsp;Zhantao Gu ,&nbsp;Ruibin Bai ,&nbsp;John Woodward ,&nbsp;Ender Özcan","doi":"10.1016/j.swevo.2025.102154","DOIUrl":"10.1016/j.swevo.2025.102154","url":null,"abstract":"<div><div>A high-quality ensemble strategy can effectively integrate several coefficients, mechanisms, and algorithms into a single framework. The adaptability, timing of intervention, and complementarity are the key factors to consider for the selected coefficients, mechanisms, and algorithms. In this study, two complementary variants based on Particle Swarm Optimization (PSO), namely Modified PSO (MPSO) and Social Learning PSO (SLPSO), were selected, forming IMPSO and ISLPSO after improvements. IMPSO excels at exploration, while ISLPSO excels at exploitation. The Improved Novel Ratio Adaptation Scheme (INRAS) is employed as a selection strategy and provides the ability to abandon less-optimal particles. The Modified Nonlinear Population Size Reduction (MNLPSR) enables the extension of generations, allowing for more sufficient evolution in later stages. Due to the use of MNLPSR, an improved inertia weight and adaptive acceleration coefficients are introduced to ensure compatibility with the proposed algorithm. Additionally, an improved dynamic differential mutation strategy is designed not only to be compatible with the proposed algorithm but also to enhance particle diversity. Both the Improved Sine Cosine Algorithm (ISCA) and Sequential Quadratic Programming (SQP), which focus on searching near the global best particles, are incorporated into the proposed ensemble strategy. This PSO-based variant is named the Effective Combination of Mechanisms for a PSO-based Ensemble Strategy (ECM-PSOES). Ablation experiments demonstrated the effectiveness of the individual coefficients and mechanisms. The novel PSO-based variant was evaluated on the CEC2017 benchmarks and compared with 14 state-of-the-art PSO-based variants and 11 non-PSO algorithms. Additionally, to evaluate the flexible and robust capability of the proposed algorithm, three real-world applications for long-term Transmission Network Expansion Planning (TNEP), Planetary Gear Train Design (PGTD), and Robot Gripper Design (RGD) were tested. The experimental results illustrate that the proposed algorithm displays superior performance compared to recently proposed PSO-based variants and most non-PSO algorithms. However, the proposed algorithm falls short of outperforming Differential Evolution (DE)-based algorithms and still requires time to match the performance of top-tier metaheuristics. The source code of ECM-PSOES is provided at <span><span>https://github.com/microhard1999/CODES</span><svg><path></path></svg></span>.</div></div>","PeriodicalId":48682,"journal":{"name":"Swarm and Evolutionary Computation","volume":"99 ","pages":"Article 102154"},"PeriodicalIF":8.5,"publicationDate":"2025-09-10","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145027492","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 multiple direction search algorithm for continuous optimization","authors":"Wei Huang ,&nbsp;Jun He ,&nbsp;Liehuang Zhu","doi":"10.1016/j.swevo.2025.102138","DOIUrl":"10.1016/j.swevo.2025.102138","url":null,"abstract":"<div><div>The particle swarm optimization algorithm has been successfully applied to various optimization problems. One of its key features is the combination of particle velocity and search direction towards the optimal position in the history and swarm. Recognizing the limitations of the particle swarm optimization algorithm, this paper proposes a new evolutionary algorithm called the multiple direction search algorithm. The algorithm integrates five different search directions, including a multi-point direction constructed using principal component analysis. The integrated direction is generated by the weighted sum of the search directions. Theoretical analysis shows that under mild conditions, the rate of convergence along the weighted direction is no worse than the rate of convergence along the best of single search directions by a positive constant, or even faster in certain cases. The performance of the proposed algorithm was evaluated on three benchmark test suites by computer simulation. Experimental results demonstrate that the proposed method outperforms seven state-of-the-art particle swarm optimization algorithms.</div></div>","PeriodicalId":48682,"journal":{"name":"Swarm and Evolutionary Computation","volume":"99 ","pages":"Article 102138"},"PeriodicalIF":8.5,"publicationDate":"2025-09-10","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145049198","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":"Constraint-tightening based adaptive two-stage evolutionary algorithm for constrained multi-objective optimization","authors":"Cunyan Liu ,&nbsp;Qingda Chen ,&nbsp;Junhua Liu ,&nbsp;Wei Zhang ,&nbsp;Meng Wang ,&nbsp;Can Liu","doi":"10.1016/j.swevo.2025.102137","DOIUrl":"10.1016/j.swevo.2025.102137","url":null,"abstract":"<div><div>Constrained multi-objective optimization problems (CMOPs) are prevalent in practical applications, yet existing methods often struggle to handle their diverse characteristics, such as disconnected feasible regions and infeasible solutions near the true constraints Pareto front (CPF). To address these challenges, this paper proposes a constraint-tightening based adaptive two-stage evolutionary algorithm (CT-TSEA) for CMOPs, incorporating a constraint boundary tightening strategy and parameter dynamic adjustment strategy. In the first stage, a constraint boundary tightening strategy based on evaluation counts guides the population toward feasible regions. Initially, constraint boundaries are relaxed to explore the solution space thoroughly, identifying promising solutions. As evaluations increase, the search boundaries shrink, enhancing the feasibility of solutions. Additionally, a step-size adaptive adjustment method improves infeasible solutions using their information, boosting search efficiency and solution diversity. The second stage introduces a dynamic adjustment method for crossover probability and scaling factor, balancing exploration and exploitation. It better balances the exploration and exploitation capabilities of the population. The proposed method is validated via comparing with seven state-of-the-art peer competitors across 59 test instances from four benchmark suites and 21 real-world problems. The corresponding results demonstrate that CT-TSEA has the higher competitiveness in addressing complex CMOPs.</div></div>","PeriodicalId":48682,"journal":{"name":"Swarm and Evolutionary Computation","volume":"99 ","pages":"Article 102137"},"PeriodicalIF":8.5,"publicationDate":"2025-09-09","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145020705","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":"Cluster and reinforcement learning-based multi-objective evolutionary algorithm for joint scheduling of virtual machines and prioritize tasks in cloud computing","authors":"Aanchal Agrawal,&nbsp;Arun Kumar Pal","doi":"10.1016/j.swevo.2025.102156","DOIUrl":"10.1016/j.swevo.2025.102156","url":null,"abstract":"<div><div>In today’s world, cloud computing is considered an essential on-demand service that is facing an ongoing problem in Virtual Machine (VM) placement and task scheduling optimization that simultaneously improves server efficiency and user experience. Considering these challenges, this paper aims to reduce the makespan, cost, and total tardiness in Joint Scheduling of Virtual Machines and Prioritize Tasks (JSVPT) by a multi-objective optimization framework. We designed a novel Cluster-Based Multi-Objective Evolutionary Algorithm (MOEA-CD/RLPD) framework, which includes a three-tier encoding scheme with Reinforcement Learning (RL)-guided local search, preselection, and dynamic resource allocation strategy to solve the problem. To guide the search process, we employ K-means clustering to decompose the population into diverse subgroups, promoting balanced exploration. The pre-selection mechanism uses a classifier to identify promising solutions in the decision space, which allows resources to be used effectively. Reinforcement learning adaptively selects intensification operators based on reward feedback, improving exploitation by intensifying promising regions of the search space. An Improved Strength Pareto Evolutionary Algorithm 2 (ISPEA2) is incorporated to maintain a diverse and high-quality Pareto archive. The performance of the proposed algorithm is assessed on multiple test instances covering different scales and benchmarked against five state-of-the-art Multi-Objective Evolutionary Algorithms (MOEAs). Experimental studies demonstrate that the proposed algorithm outperforms most existing algorithms in the literature.</div></div>","PeriodicalId":48682,"journal":{"name":"Swarm and Evolutionary Computation","volume":"99 ","pages":"Article 102156"},"PeriodicalIF":8.5,"publicationDate":"2025-09-09","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145020706","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":"DAFSP with limited assembly buffers: A deadlock-free coding-decoding paradigm and hybrid cooperative co-evolutionary approach","authors":"Siyi Wang ,&nbsp;Yanxiang Feng ,&nbsp;Xiaoling Li ,&nbsp;Guanghui Zhang","doi":"10.1016/j.swevo.2025.102155","DOIUrl":"10.1016/j.swevo.2025.102155","url":null,"abstract":"<div><div>Most prior studies on the Distributed Assembly Flowshop Scheduling Problem (DAFSP) presume infinite buffer capacity for assembly machines. However, in practical DAFSP, assembly buffers are often limited, potentially leading to a deadlock where buffers are full of jobs yet none of them can be assembled into a product. Since the deadlock in DAFSP is caused by incorrect jobs’ sequences in assembly buffers, we formulate a Petri net to model this entry process for the first time. Based on this Petri net model and improved Banker algorithm (IBA), we develop a polynomial-complexity algorithm IDAM to ensure the deadlock-free decoding of a DAFSP solution, which is coded by job and factory permutations. The makespan of such a solution is calculated backward to maintain its deadlock-free property. Furthermore, according to the proposed coding-decoding paradigm for deadlock-free solutions, we propose a hybrid cooperative co-evolution (HCCE) algorithm for DAFSP to minimize the makespan. Notably, our HCCE algorithm incorporates an elite archive (EAR) and two subpopulations. It employs problem-specific operators for heuristic initialization and global-search procedures, and four local-search operators are successively applied to every individual in the EAR. Finally, comprehensive experiments demonstrate the effectiveness and superiority of the proposed HCCE algorithm.</div></div>","PeriodicalId":48682,"journal":{"name":"Swarm and Evolutionary Computation","volume":"99 ","pages":"Article 102155"},"PeriodicalIF":8.5,"publicationDate":"2025-09-09","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145020704","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":"Dynamic performance evaluation of evolutionary multi-objective optimization algorithms for gait cycle optimization of a 25-DOFs NAO humanoid robot","authors":"Pushpendra Gupta ,&nbsp;Dilip Kumar Pratihar ,&nbsp;Kalyanmoy Deb","doi":"10.1016/j.swevo.2025.102144","DOIUrl":"10.1016/j.swevo.2025.102144","url":null,"abstract":"<div><div>Researchers are increasingly using optimization methods to achieve optimal dynamic performance of humanoid robots, often involving multiple conflicting objectives. Multi-objective optimization algorithms (MOAs) aim to find a Pareto front of optimal solutions, but selecting the best algorithm based on solution quality and computational efficiency remains challenging. This study comprehensively evaluates MOAs from different paradigms: swarm intelligence (CMOPSO), genetic algorithms (NSGA-II, DCNSGA-III), and decomposition-based approaches (CMOEA/D) for optimizing the gait cycle of a 25 DOF NAO humanoid robot during single support phase (SSP) and double support phase (DSP) scenarios. The algorithms’ convergence, diversity, and constraint-handling capabilities are systematically analyzed in solving the gait generation problem. The bi-objective optimization simultaneously minimizes power consumption and maximizes dynamic stability subject to eight functional constraints with 12-13 decision parameters. Through performance evaluation using running inverted generational distance (IGD) and hypervolume (HV) metrics across eleven independent runs of each algorithm, NSGA-II emerges as the most suitable algorithm, demonstrating superior convergence and solution quality, while CMOPSO shows competitive performance with faster initial convergence. DCNSGA-III exhibits moderate performance with constraint-handling difficulties, and CMOEA/D demonstrates poor convergence characteristics requiring significantly more computational resources. Two distinct knee regions emerge during both SSP and DSP, representing optimal trade-off solutions, with a systematic framework provided for practitioners to select appropriate gait parameters based on operational priorities. The running IGD metric combined with HV validation demonstrates effectiveness in providing robust algorithmic insights, enabling practitioners to select suitable algorithms for similar complex real-world optimization problems.</div></div>","PeriodicalId":48682,"journal":{"name":"Swarm and Evolutionary Computation","volume":"99 ","pages":"Article 102144"},"PeriodicalIF":8.5,"publicationDate":"2025-09-09","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145020720","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}