不同种类支持向量机算法的比较研究

Comparative Research on Various Support Vector Machine Algorithms

介绍一种新型的机器学习方法—支持向量机.论述了不同种类支持向量机算法并指出了每种算法的优劣.实验结果显示了核函数中选择合适的参数对分类器的效果是很重要的,通过实验还重点比较了Chunking、SMO和SVM light三种典型分解算法,并分析了训练速度优劣的原因.文章最后给出了今后SVM研究方向的一些预见.

This paper introduces a new machine learning algorithm--Support Vector Machine. This paper presents various types of support vector machine algorithms and points out their advantages and disadvantages. The experimental results show that it is important for the effect of classifier to select proper parameters of kernel function. With the comparison of the performance of the three classic SVM algorithms which are Chunking, SMO and SVMlight. The paper analyses the reasons causing the difference of training speed. In the end, the paper points out the research direction of SVM algorithms in the future.