Novel fast training algorithm for SVM based on self-adaptive steps
基于自适应步长的支持向量机快速训练算法*

The training method of SVM is to solve the convex quadratic programming. When the amount of training samples is too large, this method will not work. In order to solve this problem and improve the speed of training SVM,this paper analyzed the nature characteristics of SVM and proposed a kind of algorithm for SVM. The speed of classification was much faster than that of conventional SVM in the condition that the correct rate did not decline. The experiments on the UCI database were done with this algorithm. The experimental results show that it has better performance and partly overcomes the flaw of standard SVM, which was slow in the process of classification. This algorithm can remarkably reduce the computation and increase the speed of classification, especially in the case of large number of support vectors.