Particle swarm optimization for parameter determination and feature selection of support vector machines

摘要: Support vector machine (SVM) is a popular pattern classification method with many diverse applications. Kernel parameter setting in the SVM training procedure, along feature selection, significantly influences accuracy. This study simultaneously determines values while discovering subset of features, without reducing A particle swarm optimization (PSO) based approach for determination and selection SVM, termed PSO+SVM, developed. Several public datasets are employed to calculate accuracy rate order evaluate developed PSO+SVM approach. The was compared grid search, which conventional searching values, other approaches. Experimental results demonstrate that rates surpass those search approaches, has similar result GA+SVM. Therefore, valuable an SVM.