A novel bacterial algorithm for parameter optimization of Support Vector Machine

Support Vector Machine (SVM) plays an important role in solving the problem of data classification. However, the traditional parameter optimization method of SVM is difficult to implement and has a low precision. In order to overcome this problem, Cauchy mutation and simulated annealing (SA) algorithm are firstly applied to improve the performance of bacterial foraging optimization (BFO) algorithm. And a novel bacterial algorithm SCBFO is firstly proposed for the parameter optimization of SVM. Moreover, the validity and efficiency test of the model is strictly evaluated through learning data set the 4 UCI machine learning datasets. Three existing algorithms are employed as a contrast. The simulation results demonstrate that the proposed SCBFO-SVM method is able to obtain more accurate results.