The Research Of Classification Algorithm Based On Support Vector Machine

Classification problem is a common problem in practical applications, as well as a basic research topic in machine-learning domain. It faces many new puzzles and challenges both in theoretical research and in practical applications on the rapidly developing information technique. Support vector machine (SVM) is a novel and Powerful machine learning approach developed in the frame work of statistical learning theory, which bases on the VC theory and the Principle of structural risk minimization. It always performs well in many Practical applications with high generalization because of its better trade- off between the complexity of machines and empirical risks. Compared with traditional learning approaches, SVM holds the advantages of good generalization, being insensitive to high dimension data and convergence to global optimum, so it solves the intractable Problems of the former, such as over-learning, local minima dimension curse etc. Currently, SVM is attracting more and more researchers and becoming a new active hotspot in the fields of artificial intelligence and machine learning.This paper mainly studies the problems of classification algorithm based on support vector machine, the main contents completed as follows:1. In view of the classification of the imbalance date set, this paper gives the method using SMOTE and modify kernel. First, in date field, we use SMOTE method to processing data, reduce the imbalance. Then, in the context of algorithm, we modify support vector machine'kernel to expand the interval of small number of classes and the optimal separating hyper plane. For that we can enhance the generalization of classifier performance and the accuracy of a small number of categories and effectively solve the questions of the unbalances.2. This paper studied the multi-class SVM methods based on binary tree, propose a new multi-class SVM methods based on binary tree, this approach focused on solving the problem of division the dot and the leaf blindly and the problem of Cumulative error situation. This approach Reasonable use of the algorithm's classification of sub-classification information, greatly reduced the number of sub-classifier, thus both in time and accuracy are achieved satisfactory results.3. After studied the kernel function, we focus on talk about the problems of the basic characteristics of kernel function and how to construction a kernel function. On this basis, we propose a new method based on mixed kernel function, we use this method to deal with the imbalance data and Achieved good results.