Swarm and Evolutionary Computation最新文献_第3页

Historical knowledge transfer driven self-adaptive evolutionary multitasking algorithm with hybrid resource release for solving nonlinear equation systems 历史知识转移驱动的自适应进化多任务算法与混合资源释放，用于求解非线性方程系统

IF 8.2 1区计算机科学 Q1 COMPUTER SCIENCE, ARTIFICIAL INTELLIGENCE

Swarm and Evolutionary Computation

Pub Date : 2024-10-11 DOI: 10.1016/j.swevo.2024.101754

Yujun Zhang , Yufei Wang , Yuxin Yan , Juan Zhao , Zhengming Gao

In reality, there are many extremely complex nonlinear optimization problems. How to locate the roots of nonlinear equation systems (NESs) more accurately and efficiently has always been a major numerical challenge. Although there are many excellent algorithms to solve NESs, which are all limited by the fact that the algorithm can solve at most one NES in a single run. Therefore, this paper proposes a historical knowledge transfer driven self-adaptive evolutionary multitasking algorithm framework (EMSaRNES) with hybrid resource release to solve NESs. Its core is that in one run, EMSaRNES can efficiently and accurately locate the roots of multiple NESs. In EMSaRNES, self-adaptive parameter method is proposed to dynamically adjust parameters of the algorithm. Secondly, adaptive selection mutation mechanism with historical knowledge transfer is designed, which dynamically adjusts the evolution of populations with or without knowledge sharing according to changes in the current population diversity, thereby balancing population diversity and convergence. Finally, hybrid resource release strategy is developed, which archives the roots that meet the accuracy requirements, and then three distributions are selected to generate new populations, thus ensuring that the population diversity is maintained at high level. After a variety of experiments, it has been proven that compared to comparative algorithms EMSaRNES has superior performance on 30 general NESs test sets. In addition, the results on 18 extremely complex NESs test sets and two real-life application problems further prove that EMSaRNES finds more roots in the face of complex problems and real-life problems.

在现实生活中，存在许多极其复杂的非线性优化问题。如何更准确、更高效地找到非线性方程系统（NES）的根，一直是一个重大的数值挑战。虽然有很多优秀的算法可以求解 NES，但都受限于算法一次运行最多只能求解一个 NES。因此，本文提出了一种混合资源释放的历史知识转移驱动自适应进化多任务算法框架（EMSaRNES）来求解 NES。其核心是在一次运行中，EMSaRNES 可以高效、准确地定位多个 NES 的根。在 EMSaRNES 中，提出了自适应参数方法来动态调整算法参数。其次，设计了具有历史知识转移的自适应选择突变机制，根据当前种群多样性的变化，动态调整有无知识共享的种群进化，从而平衡种群多样性和收敛性。最后，开发了混合资源释放策略，将符合精度要求的根归档，然后选择三种分布生成新种群，从而确保种群多样性保持在较高水平。经过各种实验证明，与比较算法相比，EMSaRNES 在 30 个一般 NESs 测试集上具有更优越的性能。此外，在 18 个极其复杂的 NESs 测试集和两个实际应用问题上的结果进一步证明，EMSaRNES 在面对复杂问题和实际问题时能找到更多根源。

{"title":"Historical knowledge transfer driven self-adaptive evolutionary multitasking algorithm with hybrid resource release for solving nonlinear equation systems","authors":"Yujun Zhang , Yufei Wang , Yuxin Yan , Juan Zhao , Zhengming Gao","doi":"10.1016/j.swevo.2024.101754","DOIUrl":"10.1016/j.swevo.2024.101754","url":null,"abstract":"<div><div>In reality, there are many extremely complex nonlinear optimization problems. How to locate the roots of nonlinear equation systems (NESs) more accurately and efficiently has always been a major numerical challenge. Although there are many excellent algorithms to solve NESs, which are all limited by the fact that the algorithm can solve at most one NES in a single run. Therefore, this paper proposes a historical knowledge transfer driven self-adaptive evolutionary multitasking algorithm framework (EMSaRNES) with hybrid resource release to solve NESs. Its core is that in one run, EMSaRNES can efficiently and accurately locate the roots of multiple NESs. In EMSaRNES, self-adaptive parameter method is proposed to dynamically adjust parameters of the algorithm. Secondly, adaptive selection mutation mechanism with historical knowledge transfer is designed, which dynamically adjusts the evolution of populations with or without knowledge sharing according to changes in the current population diversity, thereby balancing population diversity and convergence. Finally, hybrid resource release strategy is developed, which archives the roots that meet the accuracy requirements, and then three distributions are selected to generate new populations, thus ensuring that the population diversity is maintained at high level. After a variety of experiments, it has been proven that compared to comparative algorithms EMSaRNES has superior performance on 30 general NESs test sets. In addition, the results on 18 extremely complex NESs test sets and two real-life application problems further prove that EMSaRNES finds more roots in the face of complex problems and real-life problems.</div></div>","PeriodicalId":48682,"journal":{"name":"Swarm and Evolutionary Computation","volume":"91 ","pages":"Article 101754"},"PeriodicalIF":8.2,"publicationDate":"2024-10-11","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142442875","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 archive-assisted multi-modal multi-objective evolutionary algorithm 档案辅助多模式多目标进化算法

IF 8.2 1区计算机科学 Q1 COMPUTER SCIENCE, ARTIFICIAL INTELLIGENCE

Swarm and Evolutionary Computation

Pub Date : 2024-10-11 DOI: 10.1016/j.swevo.2024.101738

Peng Chen , Zhimeng Li , Kangjia Qiao , P.N. Suganthan , Xuanxuan Ban , Kunjie Yu , Caitong Yue , Jing Liang

The multi-modal multi-objective optimization problems (MMOPs) pertain to characteristic of the decision space that exhibit multiple sets of Pareto optimal solutions that are either identical or similar. The resolution of these problems necessitates the utilization of optimization algorithms to locate multiple Pareto sets (PSs). However, existing multi-modal multi-objective evolutionary algorithms (MMOEAs) encounter difficulties in concurrently enhancing solution quality in both decision space and objective space. In order to deal with this predicament, this paper presents an Archive-assisted Multi-modal Multi-objective Evolutionary Algorithm, called A-MMOEA. This algorithm maintains a main population and an external archive, which is leveraged to improve the fault tolerance of individual screening. To augment the quality of solutions in the archive, an archive evolution mechanism (AEM) is formulated for updating the archive and an archive output mechanism (AOM) is used to output the final solutions. Both mechanisms incorporate a comprehensive crowding distance metric that employs objective space crowding distance to facilitate the calculation of decision space crowding distance. Besides, a data screening method is employed in the AOM to alleviate the negative impact on the final results arising from undesirable individuals resulting from diversity search. Finally, in order to enable individuals to effectively escape the limitation of niches and further enhance diversity of population, a diversity search method with level-based evolution mechanism (DSMLBEM) is proposed. The proposed algorithm’s performance is evaluated through extensive experiments conducted on two distinct test sets. Final results indicate that, in comparison to other commonly used algorithms, this approach exhibits favorable performance.

多模式多目标优化问题（MMOPs）与决策空间的特征有关，这些决策空间呈现出多组相同或相似的帕累托最优解。要解决这些问题，就必须利用优化算法来找到多个帕累托最优解集（PSs）。然而，现有的多模式多目标进化算法（MMOEAs）在同时提高决策空间和目标空间的解决方案质量方面遇到了困难。为了解决这一难题，本文提出了一种档案辅助多模式多目标进化算法，称为 A-MMOEA。该算法拥有一个主群体和一个外部档案，可用于提高单个筛选的容错性。为了提高存档中解决方案的质量，制定了存档进化机制（AEM）来更新存档，并使用存档输出机制（AOM）来输出最终解决方案。这两种机制都采用了全面的拥挤距离度量，利用目标空间拥挤距离来促进决策空间拥挤距离的计算。此外，AOM 还采用了一种数据筛选方法，以减轻多样性搜索产生的不良个体对最终结果的负面影响。最后，为了使个体有效摆脱龛位的限制，进一步提高种群的多样性，提出了一种基于水平进化机制的多样性搜索方法（DSMLBEM）。通过在两个不同的测试集上进行大量实验，对所提出算法的性能进行了评估。最终结果表明，与其他常用算法相比，该方法表现出良好的性能。

{"title":"An archive-assisted multi-modal multi-objective evolutionary algorithm","authors":"Peng Chen , Zhimeng Li , Kangjia Qiao , P.N. Suganthan , Xuanxuan Ban , Kunjie Yu , Caitong Yue , Jing Liang","doi":"10.1016/j.swevo.2024.101738","DOIUrl":"10.1016/j.swevo.2024.101738","url":null,"abstract":"<div><div>The multi-modal multi-objective optimization problems (MMOPs) pertain to characteristic of the decision space that exhibit multiple sets of Pareto optimal solutions that are either identical or similar. The resolution of these problems necessitates the utilization of optimization algorithms to locate multiple Pareto sets (PSs). However, existing multi-modal multi-objective evolutionary algorithms (MMOEAs) encounter difficulties in concurrently enhancing solution quality in both decision space and objective space. In order to deal with this predicament, this paper presents an Archive-assisted Multi-modal Multi-objective Evolutionary Algorithm, called A-MMOEA. This algorithm maintains a main population and an external archive, which is leveraged to improve the fault tolerance of individual screening. To augment the quality of solutions in the archive, an archive evolution mechanism (AEM) is formulated for updating the archive and an archive output mechanism (AOM) is used to output the final solutions. Both mechanisms incorporate a comprehensive crowding distance metric that employs objective space crowding distance to facilitate the calculation of decision space crowding distance. Besides, a data screening method is employed in the AOM to alleviate the negative impact on the final results arising from undesirable individuals resulting from diversity search. Finally, in order to enable individuals to effectively escape the limitation of niches and further enhance diversity of population, a diversity search method with level-based evolution mechanism (DSMLBEM) is proposed. The proposed algorithm’s performance is evaluated through extensive experiments conducted on two distinct test sets. Final results indicate that, in comparison to other commonly used algorithms, this approach exhibits favorable performance.</div></div>","PeriodicalId":48682,"journal":{"name":"Swarm and Evolutionary Computation","volume":"91 ","pages":"Article 101738"},"PeriodicalIF":8.2,"publicationDate":"2024-10-11","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142424805","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

Expected coordinate improvement for high-dimensional Bayesian optimization 高维贝叶斯优化的预期坐标改进

IF 8.2 1区计算机科学 Q1 COMPUTER SCIENCE, ARTIFICIAL INTELLIGENCE

Swarm and Evolutionary Computation

Pub Date : 2024-10-10 DOI: 10.1016/j.swevo.2024.101745

Dawei Zhan

Bayesian optimization (BO) algorithm is very popular for solving low-dimensional expensive optimization problems. Extending Bayesian optimization to high dimension is a meaningful but challenging task. One of the major challenges is that it is difficult to find good infill solutions as the acquisition functions are also high-dimensional. In this work, we propose the expected coordinate improvement (ECI) criterion for high-dimensional Bayesian optimization. The proposed ECI criterion measures the potential improvement we can get by moving the current best solution along one coordinate. The proposed approach selects the coordinate with the highest ECI value to refine in each iteration and covers all the coordinates gradually by iterating over the coordinates. The greatest advantage of the proposed ECI-BO (expected coordinate improvement based Bayesian optimization) algorithm over the standard BO algorithm is that the infill selection problem of the proposed algorithm is always a one-dimensional problem thus can be easily solved. Numerical experiments show that the proposed algorithm can achieve significantly better results than the standard BO algorithm and competitive results when compared with five high-dimensional BOs and six surrogate-assisted evolutionary algorithms. This work provides a simple but efficient approach for high-dimensional Bayesian optimization. A Matlab implementation of our ECI-BO is available at https://github.com/zhandawei/Expected_Coordinate_Improvement.

贝叶斯优化（BO）算法在解决低维昂贵的优化问题方面非常流行。将贝叶斯优化扩展到高维是一项有意义但极具挑战性的任务。其中一个主要挑战是，由于获取函数也是高维的，因此很难找到好的填充解决方案。在这项工作中，我们提出了用于高维贝叶斯优化的预期坐标改进（ECI）准则。所提出的 ECI 准则衡量的是我们通过沿一个坐标移动当前最佳解决方案所能获得的潜在改进。建议的方法在每次迭代中选择 ECI 值最高的坐标进行改进，并通过坐标迭代逐步覆盖所有坐标。与标准的贝叶斯优化算法相比，拟议的 ECI-BO 算法（基于预期坐标改进的贝叶斯优化算法）的最大优势在于，拟议算法的填充选择问题始终是一个一维问题，因此很容易解决。数值实验表明，所提算法的结果明显优于标准 BO 算法，与五种高维 BO 算法和六种代理辅助进化算法相比，所提算法的结果也很有竞争力。这项工作为高维贝叶斯优化提供了一种简单而高效的方法。我们的 ECI-BO 的 Matlab 实现可在 https://github.com/zhandawei/Expected_Coordinate_Improvement 上查阅。

{"title":"Expected coordinate improvement for high-dimensional Bayesian optimization","authors":"Dawei Zhan","doi":"10.1016/j.swevo.2024.101745","DOIUrl":"10.1016/j.swevo.2024.101745","url":null,"abstract":"<div><div>Bayesian optimization (BO) algorithm is very popular for solving low-dimensional expensive optimization problems. Extending Bayesian optimization to high dimension is a meaningful but challenging task. One of the major challenges is that it is difficult to find good infill solutions as the acquisition functions are also high-dimensional. In this work, we propose the expected coordinate improvement (ECI) criterion for high-dimensional Bayesian optimization. The proposed ECI criterion measures the potential improvement we can get by moving the current best solution along one coordinate. The proposed approach selects the coordinate with the highest ECI value to refine in each iteration and covers all the coordinates gradually by iterating over the coordinates. The greatest advantage of the proposed ECI-BO (expected coordinate improvement based Bayesian optimization) algorithm over the standard BO algorithm is that the infill selection problem of the proposed algorithm is always a one-dimensional problem thus can be easily solved. Numerical experiments show that the proposed algorithm can achieve significantly better results than the standard BO algorithm and competitive results when compared with five high-dimensional BOs and six surrogate-assisted evolutionary algorithms. This work provides a simple but efficient approach for high-dimensional Bayesian optimization. A Matlab implementation of our ECI-BO is available at <span><span>https://github.com/zhandawei/Expected_Coordinate_Improvement</span><svg><path></path></svg></span>.</div></div>","PeriodicalId":48682,"journal":{"name":"Swarm and Evolutionary Computation","volume":"91 ","pages":"Article 101745"},"PeriodicalIF":8.2,"publicationDate":"2024-10-10","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142424803","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

Deep reinforcement learning driven trajectory-based meta-heuristic for distributed heterogeneous flexible job shop scheduling problem 针对分布式异构灵活作业车间调度问题的基于轨迹的深度强化学习驱动元启发式

IF 8.2 1区计算机科学 Q1 COMPUTER SCIENCE, ARTIFICIAL INTELLIGENCE

Swarm and Evolutionary Computation

Pub Date : 2024-10-09 DOI: 10.1016/j.swevo.2024.101753

Qichen Zhang , Weishi Shao , Zhongshi Shao , Dechang Pi , Jiaquan Gao

As the production environment evolves, distributed manufacturing exhibits heterogeneous characteristics, including diverse machines, workers, and production processes. This paper examines a distributed heterogeneous flexible job shop scheduling problem (DHFJSP) with varying processing times. A mixed integer linear programming (MILP) model of the DHFJSP is formulated with the objective of minimizing the makespan. To solve the DHFJSP, we propose a deep Q network-aided automatic design of a variable neighborhood search algorithm (DQN-VNS). By analyzing schedules, sixty-one types of scheduling features are extracted. These features, along with six shaking strategies, are used as states and actions. A DHFJSP environment simulator is developed to train the deep Q network. The well-trained DQN then generates the shaking procedure for VNS. Additionally, a greedy initialization method is proposed to enhance the quality of the initial solution. Seven efficient critical path-based neighborhood structures with three-vector encoding scheme are introduced to improve local search. Numerical experiments on various scales of instances validate the effectiveness of the MILP model and the DQN-VNS algorithm. The results show that the DQN-VNS algorithm achieves an average relative percentage deviation (ARPD) of 3.2%, which represents an approximately 88.45% reduction compared to the best-performing algorithm among the six compared, with an ARPD of 27.7%. This significant reduction in ARPD highlights the superior stability and performance of the proposed DQN-VNS algorithm.

随着生产环境的发展，分布式制造呈现出异构特征，包括不同的机器、工人和生产流程。本文研究了处理时间不同的分布式异构灵活作业车间调度问题（DHFJSP）。本文建立了一个 DHFJSP 的混合整数线性规划（MILP）模型，目标是最小化作业时间。为了求解 DHFJSP，我们提出了一种深度 Q 网络辅助自动设计可变邻域搜索算法（DQN-VNS）。通过分析调度，我们提取了六十一种调度特征。这些特征以及六种摇摆策略被用作状态和行动。开发了一个 DHFJSP 环境模拟器来训练深度 Q 网络。训练有素的 DQN 会生成 VNS 的摇摆程序。此外，还提出了一种贪婪初始化方法，以提高初始解的质量。还引入了七种基于临界路径的高效邻域结构和三向量编码方案，以改进局部搜索。各种规模实例的数值实验验证了 MILP 模型和 DQN-VNS 算法的有效性。结果表明，DQN-VNS 算法的平均相对百分比偏差（ARPD）为 3.2%，与六种算法中表现最好的算法（ARPD 为 27.7%）相比，降低了约 88.45%。ARPD 的大幅降低凸显了拟议的 DQN-VNS 算法卓越的稳定性和性能。

{"title":"Deep reinforcement learning driven trajectory-based meta-heuristic for distributed heterogeneous flexible job shop scheduling problem","authors":"Qichen Zhang , Weishi Shao , Zhongshi Shao , Dechang Pi , Jiaquan Gao","doi":"10.1016/j.swevo.2024.101753","DOIUrl":"10.1016/j.swevo.2024.101753","url":null,"abstract":"<div><div>As the production environment evolves, distributed manufacturing exhibits heterogeneous characteristics, including diverse machines, workers, and production processes. This paper examines a distributed heterogeneous flexible job shop scheduling problem (DHFJSP) with varying processing times. A mixed integer linear programming (MILP) model of the DHFJSP is formulated with the objective of minimizing the makespan. To solve the DHFJSP, we propose a deep Q network-aided automatic design of a variable neighborhood search algorithm (DQN-VNS). By analyzing schedules, sixty-one types of scheduling features are extracted. These features, along with six shaking strategies, are used as states and actions. A DHFJSP environment simulator is developed to train the deep Q network. The well-trained DQN then generates the shaking procedure for VNS. Additionally, a greedy initialization method is proposed to enhance the quality of the initial solution. Seven efficient critical path-based neighborhood structures with three-vector encoding scheme are introduced to improve local search. Numerical experiments on various scales of instances validate the effectiveness of the MILP model and the DQN-VNS algorithm. The results show that the DQN-VNS algorithm achieves an average relative percentage deviation (ARPD) of 3.2%, which represents an approximately 88.45% reduction compared to the best-performing algorithm among the six compared, with an ARPD of 27.7%. This significant reduction in ARPD highlights the superior stability and performance of the proposed DQN-VNS algorithm.</div></div>","PeriodicalId":48682,"journal":{"name":"Swarm and Evolutionary Computation","volume":"91 ","pages":"Article 101753"},"PeriodicalIF":8.2,"publicationDate":"2024-10-09","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142424802","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 adaptive co-evolutionary competitive particle swarm optimizer for constrained multi-objective optimization problems 用于约束多目标优化问题的自适应协同进化竞争粒子群优化器

IF 8.2 1区计算机科学 Q1 COMPUTER SCIENCE, ARTIFICIAL INTELLIGENCE

Swarm and Evolutionary Computation

Pub Date : 2024-10-05 DOI: 10.1016/j.swevo.2024.101746

Xiaoding Meng , Hecheng Li

In constrained multi-objective optimization problems, it is challenging to balance the convergence, diversity and feasibility of the population, especially encountering complex infeasible regions. In order to effectively balance the three indicators, from the aspects of the handling of infeasible solution and the quality of individuals, a multi-population co-evolutionary competitive particle swarm optimization algorithm hybridized with infeasible solution transfer and an adaptive technique (ACCPSO) is proposed. Firstly, the information of feasible and infeasible individuals is fully utilized and the individuals are classified by Hamming distance. Then, a novel constraint handling technique based on learning from the promising feasible direction is designed to make individuals cross large infeasible regions and explore more potential feasible regions. Moreover, aiming to provide robust search capability and consequently further generate high-quality solutions, the genetic operators and the particle swarm optimization operator with the competitive mechanism are introduced as operators with an adaptive mechanism. Finally, compared with the state-of-the-art methods, the performance of the proposed algorithm is verified on LIR-CMOP, MW and DTLZ, as well as two real-world problems. The results indicate that ACCPSO exhibits stronger competitiveness in terms of convergence, the solution quality, and distribution diversity on the feasible Pareto front.

在受约束的多目标优化问题中，如何平衡种群的收敛性、多样性和可行性是一项挑战，尤其是在遇到复杂的不可行区域时。为了从处理不可行解和个体质量两个方面有效地平衡这三个指标，提出了一种混合了不可行解转移和自适应技术的多种群协同进化竞争粒子群优化算法（ACCPSO）。首先，充分利用可行和不可行个体的信息，通过汉明距离对个体进行分类。然后，设计了一种基于可行方向学习的新型约束处理技术，使个体跨越较大的不可行区域，探索更多潜在的可行区域。此外，为了提供稳健的搜索能力，从而进一步生成高质量的解决方案，引入了遗传算子和具有竞争机制的粒子群优化算子作为具有自适应机制的算子。最后，在 LIR-CMOP、MW 和 DTLZ 以及两个实际问题上验证了所提算法与最先进方法的性能比较。结果表明，ACCPSO 在收敛性、解的质量和可行帕累托前沿的分布多样性方面都表现出更强的竞争力。

{"title":"An adaptive co-evolutionary competitive particle swarm optimizer for constrained multi-objective optimization problems","authors":"Xiaoding Meng , Hecheng Li","doi":"10.1016/j.swevo.2024.101746","DOIUrl":"10.1016/j.swevo.2024.101746","url":null,"abstract":"<div><div>In constrained multi-objective optimization problems, it is challenging to balance the convergence, diversity and feasibility of the population, especially encountering complex infeasible regions. In order to effectively balance the three indicators, from the aspects of the handling of infeasible solution and the quality of individuals, a multi-population co-evolutionary competitive particle swarm optimization algorithm hybridized with infeasible solution transfer and an adaptive technique (ACCPSO) is proposed. Firstly, the information of feasible and infeasible individuals is fully utilized and the individuals are classified by Hamming distance. Then, a novel constraint handling technique based on learning from the promising feasible direction is designed to make individuals cross large infeasible regions and explore more potential feasible regions. Moreover, aiming to provide robust search capability and consequently further generate high-quality solutions, the genetic operators and the particle swarm optimization operator with the competitive mechanism are introduced as operators with an adaptive mechanism. Finally, compared with the state-of-the-art methods, the performance of the proposed algorithm is verified on LIR-CMOP, MW and DTLZ, as well as two real-world problems. The results indicate that ACCPSO exhibits stronger competitiveness in terms of convergence, the solution quality, and distribution diversity on the feasible Pareto front.</div></div>","PeriodicalId":48682,"journal":{"name":"Swarm and Evolutionary Computation","volume":"91 ","pages":"Article 101746"},"PeriodicalIF":8.2,"publicationDate":"2024-10-05","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142424863","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 comparative study of evolutionary algorithms and particle swarm optimization approaches for constrained multi-objective optimization problems 针对约束多目标优化问题的进化算法和粒子群优化方法的比较研究

IF 8.2 1区计算机科学 Q1 COMPUTER SCIENCE, ARTIFICIAL INTELLIGENCE

Swarm and Evolutionary Computation

Pub Date : 2024-10-03 DOI: 10.1016/j.swevo.2024.101742

Alanna McNulty , Beatrice Ombuki-Berman , Andries Engelbrecht

Many real-world optimization problems contain multiple conflicting objectives as well as additional problem constraints. These problems are referred to as constrained multi-objective optimization problems (CMOPs). Many meta-heuristics for solving CMOPs, called constrained multi-objective meta-heuristics (CMOMHs) have been introduced in the literature, including those using particle swarm optimization (PSO)(Kennedy and Eberhart, 1995), genetic algorithms (GAs)(Man et al., 1996), and differential evolution (DE)(Storn and Price, 1997). CMOMHs can be grouped into four different classes: classic CMOMHs, co-evolutionary approaches, multi-stage approaches, and multi-tasking approaches. An extensive comparative study of twenty different CMOMHs on a wide variety of test problems, including real-world CMOPs in the fields of science and engineering, is conducted. A multi-swarm PSO approach called constrained multi-guide particle swarm optimization (ConMGPSO) is introduced and compared to the best-performing previous approaches according to the comparative study. The performance of each algorithm was found to be problem dependent, however the best overall approaches were ConMGPSO, paired-offspring constrained evolutionary algorithm (POCEA)(He et al., 2021), adaptive non-dominated sorting genetic algorithm III (A-NSGA-III)(Jain and Deb, 2014), and constrained multi-objective framework using Q-learning and evolutionary multi-tasking (CMOQLMT)(Ming and Gong, 2023). ConMGPSO and POCEA had the best performance on the CF benchmark set, which contains examples of bi-objective and tri-objective CMOPs with disconnected CPOFs. The CMOQLMT approach had the best performance on the DAS-CMOP benchmark set, which contain additional difficulty in terms of feasibility-, convergence-, and diversity-hardness. For the selected real-world CMOPs, A-NSGA-III had the best performance overall. ConMGPSO was shown to have the best performance on the process, design, and synthesis problems, and had competitive performance for the power system optimization problems.

现实世界中的许多优化问题都包含多个相互冲突的目标以及额外的问题约束。这些问题被称为约束多目标优化问题（CMOPs）。文献中介绍了许多用于解决 CMOPs 的元启发式方法，称为约束多目标元启发式方法（CMOMHs），包括使用粒子群优化（PSO）（Kennedy 和 Eberhart，1995 年）、遗传算法（GAs）（Man 等人，1996 年）和微分进化（DE）（Storn 和 Price，1997 年）的方法。CMOMHs 可分为四类：经典 CMOMHs、协同进化方法、多阶段方法和多任务方法。研究人员对 20 种不同的 CMOMHs 进行了广泛的比较研究，这些 CMOMHs 应用于各种测试问题，包括科学和工程领域的实际 CMOPs。根据比较研究结果，引入了一种名为受约束多向导粒子群优化（ConMGPSO）的多粒子群 PSO 方法，并与之前表现最好的方法进行了比较。研究发现，每种算法的性能都与问题有关，但总体上最好的方法是 ConMGPSO、成对后代约束进化算法（POCEA）（He 等人，2021 年）、自适应非支配排序遗传算法 III（A-NSGA-III）（Jain 和 Deb，2014 年）以及使用 Q-learning 和进化多任务的约束多目标框架（CMOQLMT）（Ming 和 Gong，2023 年）。ConMGPSO 和 POCEA 在 CF 基准集上表现最佳，该基准集包含双目标和三目标 CMOP 示例，且 CPOF 互不关联。CMOQLMT 方法在 DAS-CMOP 基准集上表现最佳，该基准集在可行性、收敛性和多样性硬度方面存在额外困难。对于所选的真实世界 CMOP，A-NSGA-III 的总体性能最佳。ConMGPSO 在流程、设计和综合问题上表现最佳，在电力系统优化问题上也具有竞争力。

{"title":"A comparative study of evolutionary algorithms and particle swarm optimization approaches for constrained multi-objective optimization problems","authors":"Alanna McNulty , Beatrice Ombuki-Berman , Andries Engelbrecht","doi":"10.1016/j.swevo.2024.101742","DOIUrl":"10.1016/j.swevo.2024.101742","url":null,"abstract":"<div><div>Many real-world optimization problems contain multiple conflicting objectives as well as additional problem constraints. These problems are referred to as constrained multi-objective optimization problems (CMOPs). Many meta-heuristics for solving CMOPs, called constrained multi-objective meta-heuristics (CMOMHs) have been introduced in the literature, including those using particle swarm optimization (PSO)(Kennedy and Eberhart, 1995), genetic algorithms (GAs)(Man et al., 1996), and differential evolution (DE)(Storn and Price, 1997). CMOMHs can be grouped into four different classes: classic CMOMHs, co-evolutionary approaches, multi-stage approaches, and multi-tasking approaches. An extensive comparative study of twenty different CMOMHs on a wide variety of test problems, including real-world CMOPs in the fields of science and engineering, is conducted. A multi-swarm PSO approach called constrained multi-guide particle swarm optimization (ConMGPSO) is introduced and compared to the best-performing previous approaches according to the comparative study. The performance of each algorithm was found to be problem dependent, however the best overall approaches were ConMGPSO, paired-offspring constrained evolutionary algorithm (POCEA)(He et al., 2021), adaptive non-dominated sorting genetic algorithm III (A-NSGA-III)(Jain and Deb, 2014), and constrained multi-objective framework using Q-learning and evolutionary multi-tasking (CMOQLMT)(Ming and Gong, 2023). ConMGPSO and POCEA had the best performance on the CF benchmark set, which contains examples of bi-objective and tri-objective CMOPs with disconnected CPOFs. The CMOQLMT approach had the best performance on the DAS-CMOP benchmark set, which contain additional difficulty in terms of feasibility-, convergence-, and diversity-hardness. For the selected real-world CMOPs, A-NSGA-III had the best performance overall. ConMGPSO was shown to have the best performance on the process, design, and synthesis problems, and had competitive performance for the power system optimization problems.</div></div>","PeriodicalId":48682,"journal":{"name":"Swarm and Evolutionary Computation","volume":"91 ","pages":"Article 101742"},"PeriodicalIF":8.2,"publicationDate":"2024-10-03","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142424800","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

An adaptive interval many-objective evolutionary algorithm with information entropy dominance 具有信息熵优势的自适应区间多目标进化算法

IF 8.2 1区计算机科学 Q1 COMPUTER SCIENCE, ARTIFICIAL INTELLIGENCE

Swarm and Evolutionary Computation

Pub Date : 2024-10-03 DOI: 10.1016/j.swevo.2024.101749

Zhihua Cui , Conghong Qu , Zhixia Zhang , Yaqing Jin , Jianghui Cai , Wensheng Zhang , Jinjun Chen

Interval many-objective optimization problems (IMaOPs) involve more than three conflicting objectives with interval parameters. Various real-world applications under uncertainty can be modeled as IMaOPs to solve, so effectively handling IMaOPs is crucial for solving practical problems. This paper proposes an adaptive interval many-objective evolutionary algorithm with information entropy dominance (IMEA-IED) to tackle IMaOPs. Firstly, an interval dominance method based on information entropy is proposed to adaptively compare intervals. This method constructs convergence entropy and uncertainty entropy related to interval features and innovatively introduces the idea of using global information to regulate the direction of local interval comparison. Corresponding interval confidence levels are designed for different directions. Additionally, a novel niche strategy is designed through interval population partitioning. This strategy introduces a crowding distance increment for improved subpopulation comparison and employs an updated reference vector method to adjust the search regions for empty subpopulations. The IMEA-IED is compared with seven interval optimization algorithms on 60 interval test problems and a practical application. Empirical results affirm the superior performance of our proposed algorithm in tackling IMaOPs.

区间多目标优化问题（IMaOPs）涉及三个以上带有区间参数的冲突目标。不确定条件下的各种实际应用都可以建模为 IMaOPs 来解决，因此有效处理 IMaOPs 对解决实际问题至关重要。本文提出了一种具有信息熵优势的自适应区间多目标进化算法（IMEA-IED）来解决 IMaOPs。首先，本文提出了一种基于信息熵的区间优势方法，用于自适应地比较区间。该方法构建了与区间特征相关的收敛熵和不确定性熵，并创新性地引入了利用全局信息调节局部区间比较方向的思想。针对不同的方向，设计了相应的区间置信度。此外，还通过区间种群划分设计了一种新颖的利基策略。该策略引入了拥挤距离增量以改进子群比较，并采用更新的参考向量方法来调整空子群的搜索区域。在 60 个区间测试问题和一个实际应用中，IMEA-IDD 与七种区间优化算法进行了比较。实证结果表明，我们提出的算法在处理 IMaOPs 方面表现出色。

{"title":"An adaptive interval many-objective evolutionary algorithm with information entropy dominance","authors":"Zhihua Cui , Conghong Qu , Zhixia Zhang , Yaqing Jin , Jianghui Cai , Wensheng Zhang , Jinjun Chen","doi":"10.1016/j.swevo.2024.101749","DOIUrl":"10.1016/j.swevo.2024.101749","url":null,"abstract":"<div><div>Interval many-objective optimization problems (IMaOPs) involve more than three conflicting objectives with interval parameters. Various real-world applications under uncertainty can be modeled as IMaOPs to solve, so effectively handling IMaOPs is crucial for solving practical problems. This paper proposes an adaptive interval many-objective evolutionary algorithm with information entropy dominance (IMEA-IED) to tackle IMaOPs. Firstly, an interval dominance method based on information entropy is proposed to adaptively compare intervals. This method constructs convergence entropy and uncertainty entropy related to interval features and innovatively introduces the idea of using global information to regulate the direction of local interval comparison. Corresponding interval confidence levels are designed for different directions. Additionally, a novel niche strategy is designed through interval population partitioning. This strategy introduces a crowding distance increment for improved subpopulation comparison and employs an updated reference vector method to adjust the search regions for empty subpopulations. The IMEA-IED is compared with seven interval optimization algorithms on 60 interval test problems and a practical application. Empirical results affirm the superior performance of our proposed algorithm in tackling IMaOPs.</div></div>","PeriodicalId":48682,"journal":{"name":"Swarm and Evolutionary Computation","volume":"91 ","pages":"Article 101749"},"PeriodicalIF":8.2,"publicationDate":"2024-10-03","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142424801","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 dynamic-ranking-assisted co-evolutionary algorithm for constrained multimodal multi-objective optimization 约束多模态多目标优化的动态等级辅助协同进化算法

IF 8.2 1区计算机科学 Q1 COMPUTER SCIENCE, ARTIFICIAL INTELLIGENCE

Swarm and Evolutionary Computation

Pub Date : 2024-09-28 DOI: 10.1016/j.swevo.2024.101744

Guoqing Li , Weiwei Zhang , Caitong Yue , Yirui Wang , Yu Xin , Kui Gao

Constrained multimodal multi-objective optimization problems (CMMOPs) are characterized by multiple constrained Pareto sets (CPSs) sharing the same constrained Pareto front (CPF). The challenge lies in efficiently identifying equivalent CPSs while maintaining a balance among convergence, diversity, and constraints. Addressing this challenge, we propose a dynamic-ranking-based constraint handling technique implemented in a co-evolutionary algorithm, named DRCEA, specifically designed for solving CMMOPs. To search for equivalent CPSs, we introduce a co-evolutionary framework involving two populations: a convergence-first population and a constraint-first population. The co-evolutionary framework facilitates knowledge transfer and sustains diverse solutions. Subsequently, a dynamic ranking strategy is employed with dynamic weight parameters that consider both dominance and constraint relationships among individuals. Within the convergence-first population, the weight parameter for convergence gradually decreases, while the constraint parameter increases. Conversely, in the constraint-first population, the weight parameter for constraints gradually decreases, while the convergence parameter increases. This approach ensures a well-balanced consideration of convergence and constraints within the two distinct populations. Experimental results on the CMMOP test suite and the real-world CMMOP test scenario validate the effectiveness of the proposed dynamic-ranking-based constraint handling technique, demonstrating the superiority of DRCEA over seven state-of-the-art algorithms.

受限多模式多目标优化问题（CMMOPs）的特点是多个受限帕累托集（CPSs）共享同一个受限帕累托前沿（CPF）。如何在保持收敛性、多样性和约束之间的平衡的同时，高效地识别等效的 CPS 是一个挑战。为了应对这一挑战，我们提出了一种基于动态排序的约束处理技术，并将其应用于协同进化算法中，该算法被命名为 DRCEA，专门用于求解 CMMOP。为了搜索等效的 CPS，我们引入了一个共同进化框架，其中涉及两个种群：收敛优先种群和约束优先种群。共同进化框架促进了知识转移，并维持了多样化的解决方案。随后，我们采用了一种动态排序策略，其动态权重参数既考虑了个体间的支配关系，也考虑了个体间的约束关系。在收敛优先群体中，收敛权重参数会逐渐降低，而约束参数则会增加。反之，在约束优先群体中，约束的权重参数逐渐减小，而收敛的权重参数逐渐增大。这种方法确保了在两个不同的群体中对收敛和约束的均衡考虑。CMMOP 测试套件和实际 CMMOP 测试场景的实验结果验证了所提出的基于动态排序的约束处理技术的有效性，表明 DRCEA 优于七种最先进的算法。

{"title":"A dynamic-ranking-assisted co-evolutionary algorithm for constrained multimodal multi-objective optimization","authors":"Guoqing Li , Weiwei Zhang , Caitong Yue , Yirui Wang , Yu Xin , Kui Gao","doi":"10.1016/j.swevo.2024.101744","DOIUrl":"10.1016/j.swevo.2024.101744","url":null,"abstract":"<div><div>Constrained multimodal multi-objective optimization problems (CMMOPs) are characterized by multiple constrained Pareto sets (CPSs) sharing the same constrained Pareto front (CPF). The challenge lies in efficiently identifying equivalent CPSs while maintaining a balance among convergence, diversity, and constraints. Addressing this challenge, we propose a dynamic-ranking-based constraint handling technique implemented in a co-evolutionary algorithm, named DRCEA, specifically designed for solving CMMOPs. To search for equivalent CPSs, we introduce a co-evolutionary framework involving two populations: a convergence-first population and a constraint-first population. The co-evolutionary framework facilitates knowledge transfer and sustains diverse solutions. Subsequently, a dynamic ranking strategy is employed with dynamic weight parameters that consider both dominance and constraint relationships among individuals. Within the convergence-first population, the weight parameter for convergence gradually decreases, while the constraint parameter increases. Conversely, in the constraint-first population, the weight parameter for constraints gradually decreases, while the convergence parameter increases. This approach ensures a well-balanced consideration of convergence and constraints within the two distinct populations. Experimental results on the CMMOP test suite and the real-world CMMOP test scenario validate the effectiveness of the proposed dynamic-ranking-based constraint handling technique, demonstrating the superiority of DRCEA over seven state-of-the-art algorithms.</div></div>","PeriodicalId":48682,"journal":{"name":"Swarm and Evolutionary Computation","volume":"91 ","pages":"Article 101744"},"PeriodicalIF":8.2,"publicationDate":"2024-09-28","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142358431","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 leader-adaptive particle swarm optimization with dimensionality reduction strategy for feature selection 用于特征选择的领导者自适应粒子群优化与降维策略

IF 8.2 1区计算机科学 Q1 COMPUTER SCIENCE, ARTIFICIAL INTELLIGENCE

Swarm and Evolutionary Computation

Pub Date : 2024-09-27 DOI: 10.1016/j.swevo.2024.101743

Shanshan Yang , Bo Wei , Li Deng , Xiao Jin , Mingfeng Jiang , Yanrong Huang , Feng Wang

Feature selection (FS) is a key data pre-processing method in machine learning tasks. It aims to obtain better classification accuracy of an algorithm with the smallest size of selected feature subset. Particle Swarm Optimization has been widely applied in FS tasks. However, when solving FS task on high-dimensional datasets, most of the PSO-based FS methods are easy to get premature convergence and fall into the local optimum. To address this issue, a leader-adaptive particle swarm optimization with dimensionality reduction strategy (LAPSO-DR) is proposed in this paper. Firstly, a hybrid initialization strategy based on feature importance is formulated. The population is divided into two parts, which have different initialization ranges. It can not only improve the diversity of the population but also eliminate some redundant features. Secondly, the leader-adaptive strategy is proposed to improve the exploitation ability of the population, in which each particle can have a different learning exemplar selected from the elite sub-swarm. Finally, the dimensionality reduction strategy based on Markov blanket is introduced to reduce the size of the optimal feature subset. LAPSO-DR is compared with 8 representative FS methods on 18 benchmark datasets. The experimental results show that LAPSO-DR can obtain smaller sizes of feature subsets with highest classification accuracies on 17 out of 18 datasets. The classification accuracies of LAPSO-DR are over 90% on 14 datasets and the feature elimination rates are higher than 60% on 18 datasets.

特征选择（FS）是机器学习任务中的一种关键数据预处理方法。它的目的是用最小的特征子集获得算法的更高分类精度。粒子群优化（Particle Swarm Optimization）已被广泛应用于特征选择任务中。然而，当求解高维数据集上的 FS 任务时，大多数基于 PSO 的 FS 方法容易过早收敛并陷入局部最优。针对这一问题，本文提出了一种具有降维策略的领导者自适应粒子群优化（LAPSO-DR）。首先，本文提出了一种基于特征重要性的混合初始化策略。种群被分为两部分，这两部分的初始化范围不同。这不仅能提高种群的多样性，还能消除一些冗余特征。其次，为了提高种群的利用能力，提出了领导者自适应策略，即每个粒子都可以从精英子群中选择不同的学习范例。最后，引入了基于马尔可夫毯的降维策略，以减小最优特征子集的大小。在 18 个基准数据集上，LAPSO-DR 与 8 种具有代表性的 FS 方法进行了比较。实验结果表明，在 18 个数据集中的 17 个数据集上，LAPSO-DR 可以获得更小的特征子集，并获得最高的分类精度。在 14 个数据集上，LAPSO-DR 的分类准确率超过 90%，在 18 个数据集上，特征消除率超过 60%。

{"title":"A leader-adaptive particle swarm optimization with dimensionality reduction strategy for feature selection","authors":"Shanshan Yang , Bo Wei , Li Deng , Xiao Jin , Mingfeng Jiang , Yanrong Huang , Feng Wang","doi":"10.1016/j.swevo.2024.101743","DOIUrl":"10.1016/j.swevo.2024.101743","url":null,"abstract":"<div><div>Feature selection (FS) is a key data pre-processing method in machine learning tasks. It aims to obtain better classification accuracy of an algorithm with the smallest size of selected feature subset. Particle Swarm Optimization has been widely applied in FS tasks. However, when solving FS task on high-dimensional datasets, most of the PSO-based FS methods are easy to get premature convergence and fall into the local optimum. To address this issue, a leader-adaptive particle swarm optimization with dimensionality reduction strategy (LAPSO-DR) is proposed in this paper. Firstly, a hybrid initialization strategy based on feature importance is formulated. The population is divided into two parts, which have different initialization ranges. It can not only improve the diversity of the population but also eliminate some redundant features. Secondly, the leader-adaptive strategy is proposed to improve the exploitation ability of the population, in which each particle can have a different learning exemplar selected from the elite sub-swarm. Finally, the dimensionality reduction strategy based on Markov blanket is introduced to reduce the size of the optimal feature subset. LAPSO-DR is compared with 8 representative FS methods on 18 benchmark datasets. The experimental results show that LAPSO-DR can obtain smaller sizes of feature subsets with highest classification accuracies on 17 out of 18 datasets. The classification accuracies of LAPSO-DR are over 90% on 14 datasets and the feature elimination rates are higher than 60% on 18 datasets.</div></div>","PeriodicalId":48682,"journal":{"name":"Swarm and Evolutionary Computation","volume":"91 ","pages":"Article 101743"},"PeriodicalIF":8.2,"publicationDate":"2024-09-27","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142327946","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

Generating logic circuit classifiers from dendritic neural model via multi-objective optimization 通过多目标优化从树突状神经模型生成逻辑电路分类器

IF 8.2 1区计算机科学 Q1 COMPUTER SCIENCE, ARTIFICIAL INTELLIGENCE

Swarm and Evolutionary Computation

Pub Date : 2024-09-26 DOI: 10.1016/j.swevo.2024.101740

Haochang Jin , Chengtao Yang , Junkai Ji , Jin Zhou , Qiuzhen Lin , Jianqiang Li

Inspired by biological neurons, a novel dendritic neural model (DNM) was proposed in our previous research to pursue a classification technique with simpler architecture, fewer parameters, and higher computation speed. The trained DNM can be transitioned to logic circuit classifiers (LCCs) by discarding unnecessary synapses and dendrites. Unlike conventional artificial neural networks with floating-point calculations, the LCC operates entirely in binary so it can be easily implemented in hardware, which has significant advantages in dealing with a high velocity of data due to its high computational speed. However, oversimplifying the model architecture will lead to the performance degeneration of LCC, and how to balance the architecture and performance is not well understood in practical applications. Therefore, the primary motivation of this study is twofold. First, a theoretical analysis is presented that the transition of LCCs from DNM can be regarded as a specific regularization problem. Second, a multiobjective optimization framework that can simultaneously optimize the classification performance and model the complexity of LCC is proposed to solve the problem. Comprehensive experiments have been conducted to validate the effectiveness and superiority of the proposed framework.

受生物神经元的启发，我们在之前的研究中提出了一种新型树突神经模型（DNM），以追求一种结构更简单、参数更少、计算速度更快的分类技术。训练好的 DNM 可以通过舍弃不必要的突触和树突过渡到逻辑电路分类器（LCC）。与采用浮点运算的传统人工神经网络不同，逻辑电路分类器完全以二进制方式运行，因此很容易在硬件中实现，由于其计算速度快，在处理高速数据时具有显著优势。然而，过度简化模型架构会导致 LCC 性能下降，而在实际应用中如何平衡架构和性能并不十分清楚。因此，本研究的主要动机有两个方面。首先，本文从理论上分析了 LCC 从 DNM 的过渡可视为一个特定的正则化问题。其次，提出了一个多目标优化框架来解决这个问题，该框架可以同时优化分类性能和 LCC 的复杂性模型。通过综合实验验证了所提框架的有效性和优越性。

{"title":"Generating logic circuit classifiers from dendritic neural model via multi-objective optimization","authors":"Haochang Jin , Chengtao Yang , Junkai Ji , Jin Zhou , Qiuzhen Lin , Jianqiang Li","doi":"10.1016/j.swevo.2024.101740","DOIUrl":"10.1016/j.swevo.2024.101740","url":null,"abstract":"<div><div>Inspired by biological neurons, a novel dendritic neural model (DNM) was proposed in our previous research to pursue a classification technique with simpler architecture, fewer parameters, and higher computation speed. The trained DNM can be transitioned to logic circuit classifiers (LCCs) by discarding unnecessary synapses and dendrites. Unlike conventional artificial neural networks with floating-point calculations, the LCC operates entirely in binary so it can be easily implemented in hardware, which has significant advantages in dealing with a high velocity of data due to its high computational speed. However, oversimplifying the model architecture will lead to the performance degeneration of LCC, and how to balance the architecture and performance is not well understood in practical applications. Therefore, the primary motivation of this study is twofold. First, a theoretical analysis is presented that the transition of LCCs from DNM can be regarded as a specific regularization problem. Second, a multiobjective optimization framework that can simultaneously optimize the classification performance and model the complexity of LCC is proposed to solve the problem. Comprehensive experiments have been conducted to validate the effectiveness and superiority of the proposed framework.</div></div>","PeriodicalId":48682,"journal":{"name":"Swarm and Evolutionary Computation","volume":"91 ","pages":"Article 101740"},"PeriodicalIF":8.2,"publicationDate":"2024-09-26","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142323303","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