Article
Computer Science, Artificial Intelligence
Ke-Jing Du, Jian-Yu Li, Hua Wang, Jun Zhang
Summary: Evolutionary multi-objective multi-task optimization is an emerging paradigm for solving multi-objective multi-task optimization problems using evolutionary computation. This paper proposes treating these problems as multi-objective multi-criteria optimization problems and develops an algorithm framework that utilizes the knowledge of all tasks in the same population. The algorithm selects fitness evaluation functions as criteria, guided by a probability-based selection strategy and an adaptive parameter learning method. Extensive experiments show the effectiveness and efficiency of the proposed algorithm. Treating MO-MTOP as MO-MCOP is a potential and promising direction for solving these problems.
COMPLEX & INTELLIGENT SYSTEMS
(2023)
Article
Computer Science, Theory & Methods
Ye Tian, Langchun Si, Xingyi Zhang, Ran Cheng, Cheng He, Kay Chen Tan, Yaochu Jin
Summary: This article provides a comprehensive survey of state-of-the-art MOEAs for solving large-scale multi-objective optimization problems, categorizing them into different types and discussing their strengths and weaknesses. It also reviews benchmark problems for performance assessment and important applications, while also addressing remaining challenges and future research directions in evolutionary large-scale multi-objective optimization.
ACM COMPUTING SURVEYS
(2021)
Article
Computer Science, Artificial Intelligence
Elaine Guerrero-Pena, Aluizio F. R. Araujo
Summary: Dynamic multi-objective evolutionary algorithms can address multi-objective optimization problems by predicting and responding to changes, with prediction-based methods showing promise. Through the use of objective space prediction strategy and change reaction mechanism, the proposed DOSP-NSDE demonstrates competitiveness in experiments.
APPLIED SOFT COMPUTING
(2021)
Article
Computer Science, Information Systems
Chunliang Zhao, Yuren Zhou, Xinsheng Lai
Summary: This paper develops an integrated information-based evolutionary framework for solving multi-scenario multi-objective optimization problems. Two integration algorithms, scenario-based dominance principle evolutionary algorithm and decomposition-based evolutionary algorithm, are constructed using the framework, and crucial components are developed under the multi-scenario environment.
INFORMATION SCIENCES
(2022)
Article
Computer Science, Artificial Intelligence
Hui Zhang, Xiaojuan Zheng
Summary: This paper proposes a knowledge-driven adaptive evolutionary multi-objective scheduling algorithm (KAMSA) for optimizing makespan and cost of workflow execution in cloud platforms. It divides large-scale decision variables into groups using divide-and-conquer technology to improve evolutionary search efficiency. Comparison with five state-of-the-art competitors demonstrates KAMSA's advantages in 18 out of 20 test cases.
APPLIED SOFT COMPUTING
(2023)
Article
Computer Science, Artificial Intelligence
Chunliang Zhao, Yuanyuan Hao, Dunwei Gong, Junwei Du, Shujun Zhang, Zhong Li
Summary: This paper presents a general method incorporating transfer learning for multi-scenario multi-objective optimization problems (MSMOPs). It develops a multi-scenario ensemble framework that transfers knowledge between scenarios to combine arbitrary multi-objective evolutionary algorithms. An adaptive decomposition-based multi-objective evolutionary algorithm with bi-layer selection (EADaBS) is proposed and embedded within the framework as a base learner. Experimental results demonstrate the effectiveness and efficiency of the proposed algorithms, outperforming existing state-of-the-art algorithms.
KNOWLEDGE-BASED SYSTEMS
(2023)
Article
Computer Science, Information Systems
Xuefeng Hong, Mingfang Jiang, Jinglin Yu
Summary: Decomposition-based multi-objective optimization algorithms are widely used in solving complex multi-objective optimization problems. A fine-grained ensemble approach called FGEA is proposed to choose suitable evolutionary operators for different subspaces during one generation. Through an adaptive strategy, FGEA achieves competitive performance compared to five baseline algorithms on 35 complex MOPs.
Article
Computer Science, Artificial Intelligence
Yu Xue, Chen Chen, Adam Slowik
Summary: With the emergence of deep neural networks, many research fields have achieved significant breakthroughs and successfully applied them in real-life applications. In this article, a multi-objective evolutionary algorithm with a probability stack (MOEA-PS) is proposed to improve the performance of deep neural networks. The proposed algorithm considers precision and time consumption as the two objectives and uses an adjacency list to represent the internal structure of deep neural networks.
IEEE TRANSACTIONS ON EVOLUTIONARY COMPUTATION
(2023)
Review
Chemistry, Multidisciplinary
Zitong Wang, Yan Pei, Jianqiang Li
Summary: The multi-objective optimization problem is challenging due to conflicts among various objectives and functions. The research and application of multi-objective evolutionary algorithms (MOEA) have made significant progress in solving such problems. This survey provides a comprehensive investigation of MOEA algorithms, classifies them by evolutionary mechanism, and suggests the combination of chaotic evolution algorithm with representative search strategies for improving the search capability of MOEAs.
APPLIED SCIENCES-BASEL
(2023)
Article
Computer Science, Artificial Intelligence
Yinan Shao, Jerry Chun-Wei Lin, Gautam Srivastava, Dongdong Guo, Hongchun Zhang, Hu Yi, Alireza Jolfaei
Summary: This article introduces a method for optimizing deep reinforcement learning models using neural evolutionary algorithms to solve combinatorial optimization problems. The proposed end-to-end multi-objective neural evolutionary algorithm demonstrates competitive and robust performance on the classic travel salesman problem and knapsack problem, and also performs well in inference time.
IEEE TRANSACTIONS ON NEURAL NETWORKS AND LEARNING SYSTEMS
(2023)
Article
Computer Science, Artificial Intelligence
Wei Fang, Qiang Zhang, Jun Sun, Xiaojun Wu
Summary: This paper proposes a multi-objective problem model and an improved evolutionary algorithm for high quality pattern mining. Experimental results demonstrate that the proposed method outperforms existing algorithms in terms of efficiency, quality, and convergence speed.
IEEE TRANSACTIONS ON KNOWLEDGE AND DATA ENGINEERING
(2022)
Article
Computer Science, Information Systems
Boxiong Tan, Hui Ma, Yi Mei, Mengjie Zhang
Summary: As the number of functionally similar web services on the Internet continues to increase, market competition has become intense. Web service providers recognize that good Quality of Service (QoS) is crucial for business success, with low network latency being a key indicator of good QoS. This paper addresses the challenges of the Web Service Location Allocation Problem (WSLAP) by developing a new PSO-based algorithm that can provide a wider range of solutions compared to existing multi-objective optimization algorithms, particularly performing better on larger scale problems.
IEEE TRANSACTIONS ON SERVICES COMPUTING
(2021)
Article
Automation & Control Systems
Ye Tian, Haowen Chen, Haiping Ma, Xingyi Zhang, Kay Chen Tan, Yaochu Jin
Summary: In this paper, a hybrid algorithm is proposed to solve large-scale multi-objective optimization problems (LSMOPs) by combining differential evolution and conjugate gradient method. The proposed algorithm exhibits better convergence and diversity performance compared to existing evolutionary algorithms, mathematical programming methods, and hybrid algorithms on various benchmark and real-world LSMOPs.
IEEE-CAA JOURNAL OF AUTOMATICA SINICA
(2022)
Article
Automation & Control Systems
Ye Tian, Yuandong Feng, Xingyi Zhang, Changyin Sun
Summary: Evolutionary algorithms (EAs) have shown superiority in solving complex optimization problems. However, their performance deteriorates drastically when handling a large number of decision variables. To tackle this issue, a proposed efficient EA optimizes a binary vector for each solution to estimate the sparse distribution of optimal solutions and provides a fast clustering method to reduce the dimensionality of the search space. It can handle 1,000,000 real variables and reduce the runtime to less than 10% of existing EAs.
IEEE-CAA JOURNAL OF AUTOMATICA SINICA
(2023)
Article
Computer Science, Information Systems
Fernando Jimenez, Estrella Lucena-Sanchez, Gracia Sanchez, Guido Sciavicco
Summary: This paper introduces a comprehensive optimization model for detecting the causes of contamination in underground water wells, addressing the issues of selecting the best predicting variables and detecting outliers simultaneously. The results demonstrate that the proposed model can generate reliable, interpretable, and clean regression models.
Article
Engineering, Multidisciplinary
Ali Ahrari, Julian Blank, Kalyanmoy Deb, Xianren Li
Summary: This study develops a Proximity-based Surrogate-Assisted Evolutionary Algorithm (PSA-EA) for computationally expensive single-objective and multi-objective design problems. The algorithm controls the trade-off between exploration and exploitation by defining proximity and trust regions around high-fidelity solutions, aiming to maximize diversity of information in specific regions of the search space and improve the surrogate's effectiveness for future cycles simultaneously.
ENGINEERING OPTIMIZATION
(2021)
Article
Computer Science, Artificial Intelligence
Jonatas B. C. Chagas, Julian Blank, Markus Wagner, Marcone J. F. Souza, Kalyanmoy Deb
Summary: This paper proposes a method to solve a bi-objective variant of the well-studied traveling thief problem (TTP) by using a biased-random key genetic algorithm with customizations, incorporating domain knowledge, and addressing the bi-objective aspect through an elite population based on non-dominated rank and crowding distance. The method has shown success in BI-TTP competitions at EMO and GECCO conferences, consistently producing high-quality solutions.
JOURNAL OF HEURISTICS
(2021)
Article
Computer Science, Artificial Intelligence
Julian Blank, Kalyanmoy Deb, Yashesh Dhebar, Sunith Bandaru, Haitham Seada
Summary: The paper introduces a metric for generating well-spaced points on a unit simplex and proposes multiple methods for generating such a set. Through comparison on various performance metrics, the study shows that an iterative improvement based on Riesz s-energy can effectively find an arbitrary number of well-spaced points even in higher-dimensional spaces.
IEEE TRANSACTIONS ON EVOLUTIONARY COMPUTATION
(2021)
Article
Energy & Fuels
Xunzhao Yu, Ling Zhu, Yan Wang, Dimitar Filev, Xin Yao
Summary: Engine calibration is the process of optimizing engine settings to achieve optimal performance, including minimal fuel consumption, emissions, and maximum power output. With the advancement of technology, modern engines have more adjustable parameters, making the calibration task more complicated. This survey reviews the state-of-the-art applications of optimization approaches in different types of internal combustion engines, covering gasoline, diesel, and hybrid engines.
Article
Computer Science, Artificial Intelligence
Sukrit Mittal, Dhish Kumar Saxena, Kalyanmoy Deb, Erik D. Goodman
Summary: This article presents an approach that uses machine learning to learn the relationships between top solutions in optimization problems, helping offspring solutions progress. The method involves balancing tradeoffs between convergence and diversity, using the Random Forest method, and changing the application of machine learning models.
IEEE TRANSACTIONS ON EVOLUTIONARY COMPUTATION
(2022)
Article
Engineering, Multidisciplinary
Abhiroop Ghosh, Kalyanmoy Deb, Erik Goodman, Ronald Averill
Summary: This article introduces a multi-objective evolutionary algorithm framework that combines problem-specific knowledge and online innovization approaches to solve real-world large-scale multi-objective problems. The framework utilizes the knowledge of experienced users and the inter-variable relationships in good solutions to improve candidate solutions through repair operators for faster finding of good solutions.
ENGINEERING OPTIMIZATION
(2022)
Article
Engineering, Multidisciplinary
Bhuvan Khoshoo, Julian Blank, Thang Q. Pham, Kalyanmoy Deb, Shanelle N. Foster
Summary: This article investigates a complex electric machine design problem and proposes a computationally efficient optimization method based on evolutionary algorithms. The method generates feasible solutions using a repair operator and addresses time-consuming objective functions by incorporating surrogate models. The study successfully establishes the superiority of the proposed method in optimization tasks.
ENGINEERING OPTIMIZATION
(2023)
Article
Computer Science, Artificial Intelligence
Kalyanmoy Deb, Zhichao Lu, Ian Kropp, J. Sebastian Hernandez-Suarez, Rayan Hussein, Steven Miller, A. Pouyan Nejadhashemi
Summary: Many societal and industrial problems can be decomposed into hierarchical subproblems. This article introduces a new evolutionary approach that allows upper level decision makers to analyze the impact of lower level decision making when choosing a solution. This method can be applied to similar hierarchical management problems to achieve minimum deviation and more reliable outcomes.
IEEE TRANSACTIONS ON EVOLUTIONARY COMPUTATION
(2023)
Article
Computer Science, Artificial Intelligence
Dhish Kumar Saxena, Sukrit Mittal, Sarang Kapoor, Kalyanmoy Deb
Summary: This article proposes a high-fidelity-dominance principle that factors in all three critical human decision-making elements and implements it in a computationally efficient many-objective evolutionary algorithm (MaOEA). The experimental results show statistically better performance in about 60% of instances, making it practical and worthy of further investigation and application.
IEEE TRANSACTIONS ON EVOLUTIONARY COMPUTATION
(2023)
Proceedings Paper
Computer Science, Artificial Intelligence
Deepanshu Yadav, Palaniappan Ramu, Kalyanmoy Deb
Summary: Evolutionary multi-objective optimization (EMO) algorithms are commonly used to solve multi- and many-objective optimization problems and find the Pareto front. It is important for decision makers to consider objective vectors that are less sensitive to perturbations in design variables and problem parameters. This paper proposes and evaluates different algorithmic implementations that integrate multi-objective optimization, robustness consideration, and multi-criterion decision-making. The results provide insights for developing more efficient multi-objective robust optimization and decision-making procedures for practical problems with uncertainties.
PROCEEDINGS OF THE 2023 GENETIC AND EVOLUTIONARY COMPUTATION CONFERENCE, GECCO 2023
(2023)
Proceedings Paper
Computer Science, Artificial Intelligence
Ritam Guha, Wei Ao, Stephen Kelly, Vishnu Boddeti, Erik Goodman, Wolfgang Banzhaf, Kalyanmoy Deb
Summary: Automated machine learning (AutoML) greatly simplifies architecture engineering by building machine-learning algorithms using basic primitives. AutoML-Zero expands on this concept by exploring novel architectures beyond human knowledge without utilizing feature or architectural engineering. However, it currently lacks a mechanism to satisfy real-world application constraints. We propose MOAZ, a multi-objective variant of AutoML-Zero, which trades off accuracy with computational complexity, distributes solutions on a Pareto front, and efficiently explores the search space.
PROCEEDINGS OF THE 2023 GENETIC AND EVOLUTIONARY COMPUTATION CONFERENCE, GECCO 2023
(2023)
Article
Automation & Control Systems
Xunzhao Yu, Yan Wang, Ling Zhu, Dimitar Filev, Xin Yao
Summary: This article proposes a surrogate-assisted bilevel evolutionary algorithm to solve a real-world engine calibration problem. Principal component analysis is performed to investigate the impact of variables on constraints and to divide decision variables into lower-level and upper-level variables. Computational studies demonstrate that our algorithm is efficient in constraint handling and achieves a smaller fuel consumption value than other calibration methods.
IEEE TRANSACTIONS ON CYBERNETICS
(2023)
Article
Computer Science, Artificial Intelligence
Deepanshu Yadav, Palaniappan Ramu, Kalyanomy Deb
Summary: This paper proposes an approach that combines the Pareto-Race MCDM method with the interpretable self-organizing map (iSOM) based visualization method. The approach assists decision makers in multi-criteria decision-making by generating iSOM plots of objectives and considering metrics such as closeness to constraint boundaries, trade-off value, and robustness. The proposed iSOM-enabled Pareto-Race approach improves the quality of preferred solutions.
APPLIED SOFT COMPUTING
(2023)
Article
Mathematics, Interdisciplinary Applications
Kalyanmoy Deb, Matthias Ehrgott
Summary: This paper analyzes the properties of generalized dominance structures and introduces the concept of anti-dominance structure to explain the identification of resulting optimal solutions. The anti-dominance structure is applied to analyze the optimal solutions of commonly used dominance structures.
MATHEMATICAL AND COMPUTATIONAL APPLICATIONS
(2023)
Article
Automation & Control Systems
Yashesh Dhebar, Kalyanmoy Deb, Subramanya Nageshrao, Ling Zhu, Dimitar Filev, Yashesh Deepakkumar Dhebar
Summary: This article proposes a nonlinear decision-tree approach to approximate and explain the control rules of a pretrained black-box deep reinforcement learning agent. The approach uses nonlinear optimization and a hierarchical structure to find simple and interpretable rules while maintaining comparable closed-loop performance.
IEEE TRANSACTIONS ON CYBERNETICS
(2022)