Article
Automation & Control Systems
Huangke Chen, Ran Cheng, Witold Pedrycz, Yaochu Jin
Summary: This paper proposes a method to solve multiobjective optimization problems through multi-stage evolutionary search, highlighting convergence and diversity in different search stages. The algorithm balances and addresses the issues in multiobjective optimization through two stages.
IEEE TRANSACTIONS ON SYSTEMS MAN CYBERNETICS-SYSTEMS
(2021)
Article
Computer Science, Artificial Intelligence
Qiqi Liu, Yaochu Jin, Martin Heiderich, Tobias Rodemann
Summary: A new surrogate-assisted evolutionary algorithm is proposed in this study to handle expensive irregular multi-objective optimization problems. The algorithm balances convergence and diversity by adapting reference vectors and implementing a surrogate management strategy, effectively taking irregularity of the Pareto front into account.
KNOWLEDGE-BASED SYSTEMS
(2022)
Article
Computer Science, Artificial Intelligence
Amarjeet Prajapati
Summary: In this study, the performance of nine large-scale multi-objective optimization optimizers was evaluated and compared over five large-scale many-objective software clustering problems. The results showed that S3-CMA-ES and LMOSCO performed better in most cases, while H-RVEA was the worst performer.
COMPLEX & INTELLIGENT SYSTEMS
(2021)
Article
Computer Science, Artificial Intelligence
Mengzhen Wang, Fangzhen Ge, Debao Chen, Huaiyu Liu
Summary: A many-objective evolutionary algorithm with an adaptive convergence calculation method (ACC-MaOEA) is designed. It estimates the shape of the Pareto front (PF) by comparing distances and determines a reference point for convergence calculation based on PF shape. The experimental results demonstrate that ACC-MaOEA outperforms its competitors, especially on regular PF problems.
APPLIED INTELLIGENCE
(2023)
Article
Computer Science, Information Systems
Pawel B. Myszkowski, Maciej Laszczyk
Summary: The paper introduces a novel many-objective evolutionary method that aims to increase diversity and spread in the Pareto Front approximation. Experimental results show that guiding the evolution process towards less explored parts of a space can lead to increased diversity but may also increase convergence. The introduction of a novel selection operator is shown to circumvent the issue of existing diversity mechanisms in combinatorial spaces.
INFORMATION SCIENCES
(2021)
Article
Computer Science, Artificial Intelligence
Ying Xu, Huan Zhang, Xiangxiang Zeng, Yusuke Nojima
Summary: This article addresses the challenge of balancing convergence and diversity in evolutionary computation for solving many-objective optimization problems. A new preferred solution selection strategy is proposed to enhance convergence pressure by selecting non-dominated solutions with better convergence. The number of special solutions is adaptively updated to dynamically adjust the convergence pressure of the population. Experimental results demonstrate that the proposed convergence enhanced evolutionary algorithm (CEEA) outperforms state-of-the-art many-objective evolutionary algorithms.
SWARM AND EVOLUTIONARY COMPUTATION
(2022)
Article
Computer Science, Artificial Intelligence
Rui Hong, Feng Yao, Tianjun Liao, Lining Xing, Zhaoquan Cai, Feng Hou
Summary: This paper proposes a growing neural gas-assisted evolutionary many-objective optimization algorithm, GNG-EMO, which utilizes a growing neural gas network to learn the topologies of Pareto-optimal fronts and introduces an expansion mechanism for constructing better reference vectors. The performance of GNG-EMO is evaluated by comparing it with five state-of-the-art competitors in 48 test cases with irregular Pareto-optimal fronts. The numerical results show that GNG-EMO outperforms all the five competitors significantly on 27 test cases based on the indicator Hypervolume.
SWARM AND EVOLUTIONARY COMPUTATION
(2023)
Article
Computer Science, Artificial Intelligence
Yi Zhao, Jianchao Zeng, Ying Tan
Summary: The proposed method combines reference vector guided evolutionary algorithm and radial basis function networks to optimize individuals and introduces an infill strategy, showing competitive performance in solving computationally expensive many-objective optimization problems.
APPLIED SOFT COMPUTING
(2021)
Article
Computer Science, Artificial Intelligence
Hui Wang, Zichen Wei, Gan Yu, Shuai Wang, Jiali Wu, Jiawen Liu
Summary: This paper proposes a two-stage many-objective evolutionary algorithm, TS-DGPD, which accelerates convergence and maintains population diversity by using cosine distance and Lp norm, and increases selection pressure using dynamic generalized Pareto dominance. Experimental results show that the algorithm performs well in terms of convergence and diversity.
INTERNATIONAL JOURNAL OF INTELLIGENT SYSTEMS
(2022)
Article
Computer Science, Artificial Intelligence
Chunliang Zhao, Yuren Zhou, Yuanyuan Hao
Summary: This paper proposes a decomposition-based evolutionary algorithm adopting dual adjustments to address MaOPs with irregular PFs. It divides an MaOP into a set of subproblems and selects appropriate solutions using specified scalarizing functions. By updating the scalarizing functions and weight vectors, and introducing reminding solutions for fine-tuning, the algorithm performs well in experiments.
SWARM AND EVOLUTIONARY COMPUTATION
(2022)
Article
Computer Science, Artificial Intelligence
Saul Zapotecas-Martinez, Carlos A. Coello Coello, Hernan E. Aguirre, Kiyoshi Tanaka
Summary: Despite extensive studies on multi-objective test problem construction, researchers have mostly focused on designing complex search spaces, neglecting the design of Pareto optimal fronts. This paper introduces a scalable set of continuous and box-constrained multi-objective test problems, with unique Pareto fronts and features complicating the exploration of optimal solutions. The test suite provides components that can be used to construct new test instances and allows for analysis of multi-objective evolutionary algorithms.
SWARM AND EVOLUTIONARY COMPUTATION
(2023)
Article
Computer Science, Artificial Intelligence
Jun Yi, Wei Zhang, Junren Bai, Wei Zhou, Lizhong Yao
Summary: In this article, a novel MFEA based on improved dynamical decomposition (MFEA/IDD) is proposed for solving many-objective optimization problems (MaOPs). The MFEA/IDD algorithm integrates the advantages of multitasking optimization and decomposition-based evolutionary algorithms, and it effectively balances convergence and diversity while reducing the total number of function evaluations for solving MaOPs.
IEEE TRANSACTIONS ON EVOLUTIONARY COMPUTATION
(2022)
Article
Automation & Control Systems
Qiqi Liu, Yaochu Jin, Martin Heiderich, Tobias Rodemann, Guo Yu
Summary: This article proposes a new method for solving multiobjective optimization problems with irregular Pareto fronts using reference vector-based decomposition algorithm. The method uses growing neural gas network to learn the distribution of the reference vectors, achieving automatic and stable adaptation.
IEEE TRANSACTIONS ON CYBERNETICS
(2022)
Article
Computer Science, Artificial Intelligence
Chunliang Zhao, Yuren Zhou, Yuanyuan Hao, Guangyu Zhang
Summary: This paper introduces a new decomposition-based evolutionary algorithm with a bi-layer decision strategy for solving many-objective optimization problems. It accelerates convergence with adaptive fitness assignment and balances solution diversity with a diversity metric, resulting in better optimization results.
APPLIED INTELLIGENCE
(2022)
Article
Mathematics
Lining Xing, Rui Wu, Jiaxing Chen, Jun Li
Summary: This study proposes a novel many-objective evolutionary algorithm called LSEA to tackle the weakness of evolutionary many-objective algorithms based on decomposition. The LSEA performs local searches on an external archive to improve both convergence and diversity. Additionally, it perturbs the decision variables of selected solutions in order to search for better diversity and convergence. Experimental results on widely-used benchmarks demonstrate the competitive performance of the LSEA.
Article
Computer Science, Artificial Intelligence
Zhenkun Wang, Qingyan Li, Qite Yang, Hisao Ishibuchi
Summary: Dominance-resistant solutions (DRSs) are a problem in multi-objective optimization, as they can degrade the performance of multi-objective evolutionary algorithms (MOEAs). Removing DRSs can be challenging without compromising other aspects of the algorithm. This article proposes a test problem to illustrate the dilemma between eliminating DRSs and preserving boundary solutions.
COMPLEX & INTELLIGENT SYSTEMS
(2023)
Editorial Material
Computer Science, Artificial Intelligence
Jose Maria Alonso-Moral, Corrado Mencar, Hisao Ishibuchi
IEEE COMPUTATIONAL INTELLIGENCE MAGAZINE
(2022)
Article
Computer Science, Artificial Intelligence
Hisao Ishibuchi, Lie Meng Pang, Ke Shang
Summary: This paper discusses four key issues for fair comparison of multi-objective evolutionary algorithms, including termination condition, population size, performance indicators, and test problems. By analyzing the strong effects of these issues on comparison results and discussing how to address them for fair comparisons, future research directions related to each issue are suggested.
IEEE COMPUTATIONAL INTELLIGENCE MAGAZINE
(2022)
Article
Automation & Control Systems
Xiongtao Zhang, Yusuke Nojima, Hisao Ishibuchi, Wenjun Hu, Shitong Wang
Summary: This article presents a novel ensemble framework EP-TSK-FK for fuzzy subclassifiers, which achieves enhanced classification/prediction by parallel building interpretable TSK fuzzy subclassifiers and utilizing validation data. The fast classification/prediction on testing samples is achieved through iterative fuzzy C-means clustering algorithm and k-nearest neighbor (KNN) method.
IEEE TRANSACTIONS ON SYSTEMS MAN CYBERNETICS-SYSTEMS
(2022)
Article
Computer Science, Artificial Intelligence
Ke Shang, Hisao Ishibuchi, Weiyu Chen, Yang Nan, Weiduo Liao
Summary: This article investigates the hypervolume-optimal mu-distribution in three dimensions, considering line-based and plane-based Pareto fronts. It is shown that the mu solutions are not always uniformly distributed on the line-based Pareto fronts and a uniform solution set on the plane-based Pareto front is not always optimal for hypervolume maximization.
IEEE TRANSACTIONS ON EVOLUTIONARY COMPUTATION
(2022)
Article
Computer Science, Artificial Intelligence
Naoki Masuyama, Yusuke Nojima, Chu Kiong Loo, Hisao Ishibuchi
Summary: This article proposes a multi-label classification algorithm that can achieve continual learning by using an Adaptive Resonance Theory (ART)-based clustering algorithm and the Bayesian approach for label probability computation. The ART-based clustering algorithm dynamically generates prototype nodes for given data, which are used as classifiers. The label probability computation independently counts label appearances and calculates Bayesian probabilities, enabling it to handle increasing numbers of labels. Experimental results with synthetic and real-world multi-label datasets demonstrate the competitive classification performance of the proposed algorithm while maintaining continual learning capabilities.
IEEE TRANSACTIONS ON PATTERN ANALYSIS AND MACHINE INTELLIGENCE
(2023)
Article
Engineering, Electrical & Electronic
Hiroyuki Sato, Hisao Ishibuchi
Summary: Population-based evolutionary algorithms are suitable for solving multi-objective optimization problems. However, as the number of objectives increases, evolutionary multi-objective optimization faces various difficulties. This paper explains these difficulties, reviews representative approaches, and discusses their effects and limitations.
IEEJ TRANSACTIONS ON ELECTRICAL AND ELECTRONIC ENGINEERING
(2023)
Article
Computer Science, Information Systems
Ke Shang, Tianye Shu, Hisao Ishibuchi, Yang Nan, Lie Meng Pang
Summary: In the field of evolutionary multi-objective optimization, the final population is commonly used as the output, but it often contains dominated solutions from previous generations. To address this problem, a novel framework has been developed to store all non-dominated solutions in an archive and select a subset from the archive as the output. However, most studies focus on small candidate solution sets, and there is no benchmark test suite for large-scale subset selection. This study proposes a benchmark test suite and compares several subset selection algorithms using the proposed tests, providing a baseline for researchers in the EMO field to understand, compare, and develop large-scale subset selection algorithms. (c) 2022 Elsevier Inc. All rights reserved.
INFORMATION SCIENCES
(2023)
Article
Computer Science, Information Systems
Yifan Wang, Hisao Ishibuchi, Meng Joo Er, Jihua Zhu
Summary: This paper proposes an unsupervised multilayer fuzzy neural network for image clustering, which expands the applications of fuzzy systems by introducing manifold representation, calculates firing strengths using only a small number of attributes to prevent them from falling to zero, and extracts features using randomly generated convolutional weights. Experimental results demonstrate its competitiveness with existing fuzzy and nonfuzzy clustering algorithms on a wide range of image datasets.
INFORMATION SCIENCES
(2023)
Article
Computer Science, Artificial Intelligence
Jesus Guillermo Falcon-Cardona, Edgar Covantes Osuna, Carlos A. Coello Coello, Hisao Ishibuchi
Summary: In this paper, the authors investigate the use of pair-potential energy functions to improve diversity in evolutionary multi-objective optimization. They answer three important questions by developing a new algorithm for subset selection and conducting a parametrical study using a deep neural network. The results demonstrate that the utilization of pair-potential energy functions leads to good Pareto front approximations regardless of the front shape.
SWARM AND EVOLUTIONARY COMPUTATION
(2023)
Article
Computer Science, Artificial Intelligence
Linjun He, Auraham Camacho, Yang Nan, Anupam Trivedi, Hisao Ishibuchi, Dipti Srinivasan
Summary: Recently, it was found that a decomposition-based multiobjective evolutionary algorithm with a pre-specified weight vector set cannot find a uniformly-distributed solution set over an inverted triangular Pareto front. This is because the weight vectors are created by a simplex-lattice structure with a triangular shape, resulting in more boundary solutions than inside solutions. In this paper, the authors explain why the corner solutions of the inverted triangular Pareto front cannot always be found and propose a method for generating additional weight vectors to search for these solutions. Experimental results show that the proposed method improves the performance of the examined decomposition-based algorithms on multiobjective problems with irregular Pareto fronts.
SWARM AND EVOLUTIONARY COMPUTATION
(2023)
Article
Automation & Control Systems
Lie Meng Pang, Hisao Ishibuchi, Ke Shang
Summary: This article introduces the application of the multiobjective evolutionary algorithm based on decomposition (MOEA/D) with the penalty-based boundary intersection (PBI) function (MOEA/D-PBI), emphasizing the issue of penalty parameter value specification. The idea of using two different penalty parameter values simultaneously is proposed and an algorithm is designed accordingly. Experimental results demonstrate that the proposed algorithm performs well on a wide range of test problems.
IEEE TRANSACTIONS ON CYBERNETICS
(2023)
Proceedings Paper
Computer Science, Artificial Intelligence
Yuichi Omozaki, Naoki Masuyama, Yusuke Nojima, Hisao Ishibuchi
Summary: Explainable artificial intelligence (XAI) is an important research topic in machine learning. This paper introduces a fuzzy rule-based classifier as a promising XAI technique for linguistically explaining its classification result. The paper also proposes a multiobjective optimization method called multi-tasking optimization to simultaneously solve different accuracy metrics in multi-label classification. Experimental results show that the proposed method improves the accuracy of the obtained fuzzy classifiers.
2022 IEEE INTERNATIONAL CONFERENCE ON FUZZY SYSTEMS (FUZZ-IEEE)
(2022)
Proceedings Paper
Computer Science, Artificial Intelligence
Hisao Ishibuchi, Yiming Peng, Lie Meng Pang
Summary: This paper presents a new type of multi-modal multi-objective test problems, where a single point on the Pareto front corresponds to an infinite number of Pareto optimal solutions. By examining the search behavior of multi-modal multi-objective algorithms on these test problems, some interesting observations are reported.
2022 IEEE CONGRESS ON EVOLUTIONARY COMPUTATION (CEC)
(2022)
Article
Computer Science, Information Systems
Naoki Masuyama, Narito Amako, Yuna Yamada, Yusuke Nojima, Hisao Ishibuchi
Summary: This paper proposes an ART-based topological clustering algorithm that improves clustering performance by automatically estimating a similarity threshold and introduces a hierarchical structure for better information extraction.