|
Classification problem is the basic research field of machine learning problems , it is widely used in practice . Naive Bayesian classifier is a simple and effective classifier, which is based on the assumption of independence between attributes . Based on statistical learning theory support vector machines , in dealing with small samples , nonlinear and high dimensional pattern recognition problem showing some unique advantages . This article briefly describes about the naive Bayes classifier and support vector machine 's basic theory, including Naive Bayesian classification model , the basic problems of machine learning , statistical learning theory, basic content , support vector machine classification algorithm , Copula function principles and nature . In this paper, Naive Bayes classifier and support vector machine classification proposed two improved algorithms. For naive Bayes classifier attributes the lack of independence assumption , made use of the theory of Copula Copula Based Bayesian classification algorithm , further extends the naive Bayes classifier , the experimental results show that the Bayesian Copula classification algorithm achieved a better classification results. This article also constructed Copula function based on Mahalanobis distance and the distance mapping and conditional probability mapping, a sample of high- dimensional space into two-dimensional space of the new sample , and new samples for easy handling , constructed separable support vector sorting machines , which is characterized by a simple algorithm, easy to implement. Simulation results show that the structure of the classification algorithm achieved good results. Finally, a summary of the full text of the work , and pointed out the problems to be studied further .
|