Article
Computer Science, Information Systems
Jakub Kudela, Radomil Matousek
Summary: Standard evolutionary optimization algorithms assume that the evaluation of objective and constraint functions is simple and inexpensive, but in many real-world problems, these evaluations are computationally expensive. Surrogate-assisted evolutionary algorithms (SAEAs) integrate an evolutionary algorithm with a surrogate model that approximates the expensive function. This paper proposes a surrogate model based on Lipschitz underestimation and develops a differential evolution-based algorithm called LSADE, which performs competitively compared to state-of-the-art algorithms, especially for high-dimensional complex benchmark functions.
INFORMATION SCIENCES
(2023)
Article
Computer Science, Artificial Intelligence
Xinfang Ji, Yong Zhang, Dunwei Gong, Xiaoyan Sun
Summary: This article proposes a dual-surrogate-assisted cooperative particle swarm optimization algorithm for expensive multimodal optimization problems, combining dual-population cooperative particle swarm optimizer and modal-guided dual-layer cooperative surrogate model, with a hybrid strategy for detecting new modalities. Experimental results show that the algorithm can obtain multiple highly competitive optimal solutions at a low computational cost.
IEEE TRANSACTIONS ON EVOLUTIONARY COMPUTATION
(2021)
Article
Computer Science, Artificial Intelligence
Qiuzhen Lin, Xunfeng Wu, Lijia Ma, Jianqiang Li, Maoguo Gong, Carlos A. Coello Coello
Summary: This article proposes an ensemble surrogate-based framework for solving computationally expensive multiobjective optimization problems (EMOPs). The framework trains a global surrogate model and multiple surrogate submodels to enhance prediction accuracy and reliability. Experimental results demonstrate the advantages of this approach in solving EMOPs.
IEEE TRANSACTIONS ON EVOLUTIONARY COMPUTATION
(2022)
Review
Computer Science, Artificial Intelligence
Chunlin He, Yong Zhang, Dunwei Gong, Xinfang Ji
Summary: This paper provides a systematic overview of surrogate-assisted evolutionary algorithms (SAEAs), including the necessity of studying SAEAs, commonly used surrogate models, classification and discussion of existing SAEAs, review of their applications in various fields, and suggestions for future research directions.
EXPERT SYSTEMS WITH APPLICATIONS
(2023)
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
Automation & Control Systems
Xunfeng Wu, Qiuzhen Lin, Jianqiang Li, Kay Chen Tan, Victor C. M. Leung
Summary: In this article, an ensemble surrogate-based coevolutionary optimizer is proposed to solve large-scale optimization problems. By training local surrogate models and using feature selection to construct a selective ensemble surrogate, the optimizer approximates the target problem. With two populations solving the target problem and a simplified auxiliary problem collaboratively, the coevolutionary optimizer can leverage the search experience from the auxiliary problem to help solve the target problem.
IEEE TRANSACTIONS ON CYBERNETICS
(2023)
Article
Automation & Control Systems
Jianping Luo, Liang Chen, Xia Li, Qingfu Zhang
Summary: This study investigates multitask learning using conditional neural process networks and proposes two multitask learning models based on CNPs. Compared with existing models, the proposed models improve performance by adding a correlation learning layer to learn the correlation among tasks. Moreover, the proposed Bayesian optimization framework utilizes the possible dependencies among tasks to share knowledge and the proposed surrogate models confidently estimate model parameters. Experimental results show that the proposed algorithms are competitive in performance compared with other model-based optimization methods.
IEEE TRANSACTIONS ON CYBERNETICS
(2022)
Article
Computer Science, Information Systems
Jeng-Shyang Pan, Nengxian Liu, Shu-Chuan Chu, Taotao Lai
Summary: Surrogate-assisted evolutionary algorithms (SAEAs) combine the searching capabilities of evolutionary algorithms with the predictive capabilities of surrogate models, and an efficient SAHO algorithm integrates TLBO and DE algorithms, alternating between global exploration and local exploitation when better solutions cannot be found, with a new prescreening criterion selecting promising candidates for evaluations, and using a local RBF surrogate model to mimic the target function landscape.
INFORMATION SCIENCES
(2021)
Article
Computer Science, Artificial Intelligence
Yong Wang, Jianqing Lin, Jiao Liu, Guangyong Sun, Tong Pang
Summary: This article proposes a surrogate-assisted differential evolution algorithm with region division (ReDSADE) to solve expensive optimization problems with discontinuous objective functions. The algorithm combines region division, Kriging-based search, and RBF-based local search strategies, achieving good convergence accuracy and speed.
IEEE TRANSACTIONS ON EVOLUTIONARY COMPUTATION
(2022)
Article
Computer Science, Artificial Intelligence
Mingyuan Yu, Jing Liang, Kai Zhao, Zhou Wu
Summary: Surrogate-assisted evolutionary algorithms have received increasing attention in solving computationally expensive engineering optimization problems. This paper proposes a novel model management strategy based on multi-RBF parallel modeling technology to adaptively select a high-fidelity surrogate during the optimization process. Experimental results show that the proposed algorithm is robust and efficient.
SWARM AND EVOLUTIONARY COMPUTATION
(2022)
Article
Automation & Control Systems
Jianping Luo, YongFei Dong, Zexuan Zhu, Wenming Cao, Xia Li
Summary: This study proposes a surrogate methodology based on information transfer to improve the estimation effectiveness of surrogate models in multiobjective optimization problems. By mapping related tasks and training a multitask Gaussian process model, the confidence in parameter learning is enhanced, and the predicted values of objective functions can be obtained through reverse mapping. Experimental tests demonstrate that this approach outperforms other surrogate-based optimization algorithms.
IEEE TRANSACTIONS ON SYSTEMS MAN CYBERNETICS-SYSTEMS
(2023)
Article
Engineering, Multidisciplinary
Zachary Cosenza, David E. Block
Summary: The study introduced a hybrid surrogate framework combining radial basis function/genetic algorithm, which showed at least as good performance as its constituent algorithms in high-dimensional test functions, making it practical for various optimization design problems. Experiments demonstrated that the framework could be further enhanced for processes with simulated noise.
ENGINEERING OPTIMIZATION
(2021)
Article
Computer Science, Artificial Intelligence
Zhihai Ren, Chaoli Sun, Ying Tan, Guochen Zhang, Shufen Qin
Summary: The proposed bi-stage surrogate-assisted hybrid algorithm utilizes global and local searches to find optimal solutions in high-dimensional problems, demonstrating better performance compared to state-of-the-art methods for solving expensive problems.
COMPLEX & INTELLIGENT SYSTEMS
(2021)
Article
Computer Science, Interdisciplinary Applications
Qi Zhou, Jinhong Wu, Tao Xue, Peng Jin
Summary: The paper introduces a two-stage adaptive multi-fidelity surrogate model-assisted multi-objective genetic algorithm (AMFS-MOGA), which involves obtaining a preliminary Pareto frontier using low-fidelity model data in the first stage and constructing an initial MFS model based on samples selected from the preliminary Pareto set in the second stage. The fitness values of individuals are evaluated using the MFS model, which is adaptively updated according to prediction uncertainty and population diversity. The effectiveness of the proposed approach is demonstrated through benchmark tests and design optimization, showing comparable results to traditional methods while significantly reducing computational costs.
ENGINEERING WITH COMPUTERS
(2021)
Article
Computer Science, Artificial Intelligence
Qinghua Gu, Qian Wang, Xuexian Li, Xinhong Li
Summary: A new algorithm, RFMOPSO, is proposed in this paper to optimize constrained combinatorial optimization problems by combining multi-objective particle swarm optimization with a random forest model. Adaptive ranking strategy and novel rule are employed to improve search speed and adaptively update particle states for better objective balance and feasible solutions. Experimental results show promising performance on benchmark problems with discrete variables varying from 10 to 100.
KNOWLEDGE-BASED SYSTEMS
(2021)