Machine training and parameter settings with social emotional optimization algorithm for support vector machine

Abstract

Machine training along with the parameter settings significantly influences the performance of support vector machine (SVM). In this paper, the social emotional optimization algorithm (SEOA) characterized by excellent global optimization ability is employed for machine training and parameter settings for SVM. Instead of the quadratic programming problem, machine training for SVM is modeled as a multi-parameter optimization problem which is solved by SEOA. Besides, SEOA is also employed for SVM parameter settings. The kernel function parameter and error penalty parameter of SVM are simultaneously optimized by SEOA. The experiments adopt several real world datasets from the UCI database. The results indicate that training SVM with SEOA is feasible and effective. The trained SVM can achieve high classification accuracy while using fewer support vectors. Compared with cross validation method and PSO, SEOA is higher efficient in parameter settings of SVM.