.png){kind=icon}
返回
Dynamic Multivariation Multifactorial Evolutionary Algorithm for Large-Scale Multiobjective Optimization
.png){kind=icon}
PRE
AI
DOI：10.1109/TETCI.2025.3597299.png)
.png){kind=icon}
原文求助.png){kind=icon}
分享.png){kind=icon}
收藏
摘要
En 中文
Problem transformation-based multiobjective evolutionary algorithms (MOEAs) face the risk of losing optimal solutions when transforming a large-scale multiobjective optimization problem into a low-dimensional variant. One possible solution is to conduct concurrent multivariation searches, where the original space is treated as the original task, and the transformed spaces are treated as auxiliary tasks. However, determining reasonable dimensions for these transformed spaces remains a challenge. To address this, we propose a novel multivariation multifactorial evolutionary algorithm that dynamically adjusts the auxiliary tasks, with each corresponding to a transformed space with an optimal dimension for the current evolutionary stage. The approach begins by initializing a pool of candidate tasks with varying dimensions, allowing them to participate in the evolutionary process. It then quantifies each candidate's contribution and selects the most efficient ones as auxiliary tasks at each evolutionary stage. Additionally, we propose a novel knowledge transfer strategy that enables accurate cross-task and cross-dimensional transfer by integrating high-dimensional and low-dimensional information, thereby enhancing information sharing efficiency among multiple optimization tasks. Experimental results from a benchmark test suite and various real-world problems indicate the superior performance of the proposed algorithm in the majority of cases, demonstrating strong competitiveness when compared to state-of-the-art MOEAs.
Keyword:
Optimization
Convergence
Search problems
Heuristic algorithms
Knowledge transfer
Accuracy
Transforms
Training
Particle swarm optimization
Information sharing
Large-scale multiobjective optimization
multiobjective evolutionary algorithms
multifactorial optimization
dimension of auxiliary task
期刊
I
IF:
6.5
论文数:
1.4K
被引数:
4.5K
机构
引用论文
.png){kind=icon}
PlatEMO: A MATLAB Platform for Evolutionary Multi-Objective OptimizationPlatEMO: 用于进化多目标优化的MATLAB平台
OAAI
OAAI分享收藏A Survey on Learnable Evolutionary Algorithms for Scalable Multiobjective Optimization可扩展多目标优化的可学习进化算法综述
OAAI
OAAI分享收藏分享收藏A random dynamic grouping based weight optimization framework for large-scale multi-objective optimization problems基于随机动态分组的大规模多目标优化问题权重优化框架
PREAI
分享收藏
查看更多内容