A Dimension-Wise Particle Swarm Optimization Algorithm Optimized via Self-Tuning

摘要

This article proposes an improvement to the traditional Particle Swarm Optimization (PSO) via modifications w.r.t. how particles move and are attracted to optimal positions. The performance is evaluated based on how well the algorithm is able to perform w.r.t. finding the global maxima of the Sinc, Marr-Wavelet, and Drop-Wave functions in multi-dimensional problem spaces. Each algorithm is put through a session of self-tuning with a sufficient number of iterations to ensure convergence so as to demonstrate that the evaluation of each algorithm is done with justified optimal parameters.