The classification of breast tumors based on the improved PSO-SVM

摘要

The support vector machine (SVM) is an extensively used machine learning method in many biomedical signal classification applications. In this study, an improved PSO-SVM model is utilized to improve breast tumor classification accuracy. This method hybridizes the improved particle swarm optimization (PSO) and the SVM. This optimization mechanism involves kernel parameter setting of the SVM through the PSO algorithm. But, due to the shortage of the PSO algorithm itself, this paper uses the improved PSO algorithm. Through promoting the learning factor of basic particle swarm optimization, this algorithm gets more accurate parameter values to minimize SVM prediction error and raises breast tumor classification accuracy. Firstly, we extract some important indicators of breast tumor, normalize these indicators' data, and then use the improved PSO algorithm to optimize the parameters of the SVM model, finally compare the PSO-SVM method and the SVM method with the improved PSO-SVM method. The experimental results show that the improved PSO-SVM algorithm has high classification accuracy.