Research Of Support Vector Machine Learning Algorithms

Support Vector Machine (SVM), which is one implementation in Statistical Learning Theory, has some advantages, such as Global optimum, simple structure and good generalization, it has been applied to many pattern recognition and machine learning fields successfully and become a popular ongoing research area. However, its need for complexity of computation and storage requirements is the bottle-neck to deal with large-scale data. To overcome the disadvantages of SVM in training speed and precision, some researches are carried out aimed at incremental learning and multi-class classification based on SVM in this paper.This paper's basic concepts of Statistical Learning Theory and SVM are summarized firstly, which are the groundwork of next research works.For the study of SVM training algorithm, training a Support Vector Machines the solution of a very large Quadratic Programming(QP) optimization problem. Traditional optimization methods cannot be directly applied due to memory restrictions, especially for lager-scale samples. So, designing a effective SVM training method is becoming a important content of the research of SVM. Up to now, several approaches exist for circumventing the above shortcomings and work well. The dissertation explores a new SVM training algorithm-SBA based on density of samples. It mainly using the introduction of density function of sample density, and choose the initial set and the new SBA based on suitable density, the algorithm can speed up the convergence rate.Finally,the disadvantages of the existing methods of Support Vector Machine multi-class classification are analyzed and compared in this paper, such as 1-a-r is difficult to train and the classifying speed of 1-a-1 is slow. To solve these problems, a multi-class Support Vector Machine classification Algorithm based on SBA and KNN is proposed in this paper. It mainly to apply the SBA algorithm to the binary tree of multi-classification algorithm, and improve the choosing method of the initial set of options, can shorten training time while in the process of integration into the KNN classification algorithm to improve classification accuracy.In this paper SVM are combined with other algorithms and some new algorithms are proposed by making good use of SVM classifies itself and the advantages of other efficient algorithms. Some simulation experiments are carried out on software platform of MATLAB 7.0 and experiment results are analyzed and summarized at last.