A novel context-free grammar for the generation of PSO algorithms

Abstract

Particle swarm optimization algorithm (PSO) has been widely studied over the years due to its competitive results in different applications. However, its performance is dependent on some design components (e.g., inertia factor, velocity equation, topology). Thus, to define which is the best algorithm design to solve a given optimization problem is difficult due to the large number of variations and parameters that can be considered. This work proposes a novel context-free grammar for Grammar-Guided Genetic Programming (GGGP) algorithms to guide the creation of Particle Swarm Optimizers. The proposed grammar considers four aspects of the PSO algorithm that may strongly impact on its performance: swarm initialization, neighborhood topology, velocity update equation and mutation operator. To assess the proposal, a GGGP algorithm was set with the proposed grammar and employed to optimize the PSO algorithm in 32 unconstrained continuous optimization problems. In the experiments, we compared the algorithms generated from the proposed grammar with those algorithms produced by two other grammars presented in the literature to automate PSO designs. The results achieved by the proposed grammar were better than the counterparts. Besides, we also compared the generated algorithms to 6 competition algorithms with different strategies. The experiments have shown that the algorithms generated from the grammar reached better results.