A Particle Swarm Optimizer based on spiking oscillators with dynamic thresholds for velocity vectors

Deterministic Particle Swarm Optimization (DPSO) variants, while useful for analyzing PSO behavior, often struggle with exploration-exploitation balance in complex landscapes. This paper introduces an improved DPSO variant based on the spiking oscillator model. The proposed algorithm employs a modified position update mechanism using the particle's velocity components, rather than the position components as in previous algorithm. This modification enhances search space exploration without reliance on random number generators. We evaluated on the CEC 2013 benchmark functions and compared with previous algorithms which uses the position components as a threshold condition. The proposed algorithm outperforms the previous one in complex landscapes, especially in the composition functions of the benchmark functions, demonstrating improved efficiency in complex optimization tasks.