LEM-PSO: a lightweight evolutionary-state-driven multiple information learning particle swarm optimization algorithm

• Published in: Neural computing & applications Vol. 37; no. 27; pp. 22667 - 22688
• Main Authors: Yang, Xu, Li, Hongru
• Format: Journal Article
• Language: English
• Published: London Springer London 01.09.2025; Springer Nature B.V
• Subjects: Accuracy; Algorithms; Artificial Intelligence; Computational Biology/Bioinformatics; Computational Science and Engineering; Computer Science; Data Mining and Knowledge Discovery; Efficiency; Fitness; Genetic algorithms; Image Processing and Computer Vision; Machine learning; Neighborhoods; Optimization; Particle swarm optimization; Probability and Statistics in Computer Science; S.I.: Hybrid Approaches to Nature-inspired Optimization Algorithms and Their Applications; Special Issue on Hybrid Approaches to Nature-inspired Optimization Algorithms and Their Applications; Diversity information; Lightweight; Evolutionary state; Particle swarm optimization
• ISSN: 0941-0643; 1433-3058
• DOI: 10.1007/s00521-025-11083-y

Particle swarm optimization (PSO) has been widely used, in which each particle selects its learning sample relying on fitness information. Intuitively, fitness-based selection strategy is beneficial to optimization. However, excessive reliance on fitness information may cause premature convergence of the whole population. To solve the defects of PSO, a lightweight evolutionary-state-driven multiple information learning particle swarm optimization algorithm (LEM-PSO) is proposed. In the new proposed LEM-PSO, firstly, a lightweight multiple information learning strategy is proposed. Then, adaptive evolutionary-state adjustment mechanism is proposed. Finally, local optimum warning operation is used to help the stagnant population to jump from local optimums. The comprehensive performance of LEM-PSO is compared with seven popular PSO variants on CEC2013, CEC2017 and two engineering problems, and the results confirm the firmness of LEM-PSO.