Some modifications in binary particle swarm optimization for dimensionality reduction

Abstract

Binary particle swarm optimization (BPSO) algorithm is a nature inspired algorithm which has seen many applications in optimization across disciplines. BPSO algorithm consists of mainly equations for velocity updation and position updation. In this paper some of the modifications for binary particle swarm optimization (PSO) on those equations are suggested. The classification results of benchmarking dataset, namely, SRBCT and DLBCL with feature selection using proposed modifications and improved binary PSO (IBPSO) show that they have equal merits.