Support Vector Machine Applied Research In The Field Of Pattern Recognition

Support Vector Machine is a novel powerful machine learning method developed in the framework of Statistical Learning Theory(SLT) which provided by Vapink.It uses Structure Risk Minimization(SRM) and Vapnik-Chervonenkis Dimension(VC) as theory base,minimized the Practical Risk of learning machine by select function subsets and its decision function correctly,it make sure to gained minimum error classifier by limited samples.SVM solves practical problems such as small samples,nonlinearity,over learning,high dimension and local minima,which exit in most of learning methods,at the same time,it has better generalization capability.Currently,being the optimal learning theory for small samples,SLT and SVM is attracting more and more researcher and becoming a new active area in the field of Artificial Intelligent and Machine Learning.The main research of this paper can be classed as follows:improved algorithm of Support Vector Machine,multi-classification algorithm of SVM,study of Support Vector Machine algorithm for Pattern Recognition and so on.The main results of the paper are as follows:(1)For the bottle-neck problem of original incremental algorithm of Support Vector Machine,this paper presents a new incremental algorithm of Support Vector Machine, which makes the reserved samples are all the most effective by adds KKT(Karush-Kuhn-Tucke) condition as the restrict condition and threshold the decision function,improve the operation speed and classification efficiency.(2)For the original multi-classification SVM which is based on SVM-decision binary tree has too many training samples and high time expended,the paper presents a multi-classification SVM which bases on clustering idea.It improves the operation speed by import Spatial Distance and Clustering idea.The experiment result indicated that the new algorithm not only can induce the time expended,but also can keep the generalization capability,improve the operation speed and classification efficiency.(3)Introduce the Pattern Recognition theory systematically,and apply the improved algorithms to the Ship Recognition and Face Recognition,Compare with the original multi-classification SVM algorithms,the improved algorithm has obvious advantages on the operation speed and classification efficiency.