An experimental approach to design deterministic and adaptive control schemes for grouping genetic algorithms

Genetic algorithms can solve many complex problems, including designing and optimizing machine learning techniques like neural networks, as well as challenges in production management and engineering. Paradoxically, the design of these methods, which aim to solve optimization problems efficiently, d...

Full description

Saved in:
Bibliographic Details
Published in:Neural computing & applications Vol. 37; no. 33; pp. 27811 - 27840
Main Authors: Flores-Torres, Leonardo, Quiroz-Castellanos, Marcela, Ramos-Figueroa, Octavio, Amador-Larrea, Stephanie
Format: Journal Article
Language:English
Published: London Springer London 01.11.2025
Springer Nature B.V
Subjects:
Adaptive control
Artificial Intelligence
Combinatorial analysis
Computational Biology/Bioinformatics
Computational Science and Engineering
Computer Science
Configuration management
Data Mining and Knowledge Discovery
Design
Exploitation
Genetic algorithms
Heuristic methods
Image Processing and Computer Vision
Machine learning
Mutation
Neural networks
Optimization
Parameter identification
Probability and Statistics in Computer Science
Production management
S.I.: 2023 India International Congress on Computational Intelligence
Solution space
Space exploration
Special Issue on 2023 India International Congress on Computational Intelligence
Grouping genetic algorithm
Parameter setting
Parameter control
Adaptive control
Deterministic control
ISSN:0941-0643, 1433-3058
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:Genetic algorithms can solve many complex problems, including designing and optimizing machine learning techniques like neural networks, as well as challenges in production management and engineering. Paradoxically, the design of these methods, which aim to solve optimization problems efficiently, depends, in turn, on their components’ design and optimal configuration. Grouping genetic algorithms (GGAs) have excelled in performance and adaptability as one of the best metaheuristics for solving combinatorial optimization problems that require finding optimal partitions of sets of items; however, their performance relies heavily on the proper configuration of parameters like population size, crossover rate, and mutation rate. This paper presents an experimental approach for automated parameter control of GGAs, looking for a dynamic adjustment, enhancing the algorithm’s ability to explore the solution space efficiently, avoid premature convergence, and improve overall solution quality. A comprehensive set of deterministic and adaptive control schemes is introduced for on-line parameter setting in GGAs. The approach is tested by studying three state-of-the-art algorithms for solving complex instances of three NP-hard optimization problems: the Grouping Genetic Algorithm with Controlled Gene Transmission for the One-Dimensional Bin Packing Problem, the Grouping Genetic Algorithm with Intelligent Heuristic Strategies for the Parallel-Machine Scheduling Problem with Unrelated Machines and Makespan Minimization, and the Grouping Genetic Algorithm for Variable Decomposition in Large-Scale Constrained Optimization Problems. The experimental results showed that the proposed approach allows for identifying parameter control schemes that save the extensive task of off-line parameter fine-tuning, obtaining a robust and competitive performance on different benchmark sets, and outperforming the published results for some classes of instances.
Bibliography:ObjectType-Article-1
SourceType-Scholarly Journals-1
ObjectType-Feature-2
content type line 14
ISSN:0941-0643
1433-3058
DOI:10.1007/s00521-025-11270-x