A clustering-based coevolutionary multi-objective evolutionary algorithm for handling robust and noisy optimization

The presence of uncertainty is commonplace in real-world scenarios. Uncertainties can be present in both the objective space and the decision space in optimization problems. These uncertainties can pose significant challenges for evolutionary algorithms. For example, perturbations in decision variab...

Full description

Saved in:
Bibliographic Details
Published in:Evolutionary intelligence Vol. 17; no. 5-6; pp. 3767 - 3791
Main Authors: de Sousa, Mateus Clemente, Meneghini, Ivan Reinaldo, Guimarães, Frederico Gadelha
Format: Journal Article
Language:English
Published: Berlin/Heidelberg Springer Berlin Heidelberg 01.10.2024
Springer Nature B.V
Subjects:
Applications of Mathematics
Artificial Intelligence
Bioinformatics
Clustering
Control
Decision making
Engineering
Evolutionary algorithms
Function generators
Genetic algorithms
Mathematical and Computational Engineering
Mechatronics
Multiple objective analysis
Noise levels
Optimization
Pareto optimum
Research Paper
Robotics
Robustness (mathematics)
Statistical Physics and Dynamical Systems
Uncertainty
Variables
Multi-objective optimization
Robust optimization
Noisy optimization
Clustering
Evolutionary algorithm
ISSN:1864-5909, 1864-5917
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:The presence of uncertainty is commonplace in real-world scenarios. Uncertainties can be present in both the objective space and the decision space in optimization problems. These uncertainties can pose significant challenges for evolutionary algorithms. For example, perturbations in decision variables (Robust Optimization) and noise in objective functions (Noisy Optimization). Despite the plethora of methods proposed for Robust or Noisy Optimization, addressing both forms of uncertainty concurrently remains an open research question. We introduce a novel approach based on TEDA-CMOEA/D, augmented with clustering techniques for descendant generation in Robust and Noisy Optimization problems. Notably, the proposed algorithm yields promising results for uncertainty simultaneously sans the requirement for sampling, thereby reducing computational complexity. We leverage an extension of an existing test function generator for Multi-Objective Optimization of the tests. The benchmark integrates uncertainties in decision variables and/or objective functions. Experimental evaluations encompassed varying noise intensities, elucidating the impact of different noise levels on algorithmic performance. The results demonstrate the superior performance of the proposed approach compared to existing algorithms, specifically RNSGA-II and CRMOEA/D. The proposed algorithm emerges as a promising solution for Robust and Noisy Multi-Objective Optimization problems.
Bibliography:ObjectType-Article-1
SourceType-Scholarly Journals-1
ObjectType-Feature-2
content type line 14
ISSN:1864-5909
1864-5917
DOI:10.1007/s12065-024-00956-1