Study Of Several Optimization Algorithms For Support Vector Machine

Support Vector Machine (SVM) is a novel machine learning method and has been successfully applied into the field of data classification and regression such as handwritten digit recognition, face recognition, text classification, regression prediction and time serial analysis etc. How to design the efficient optimization algorithms for solving SVM has become an open problem. In this paper, based on the optimization algorithms for solving SVM which is proposed by O.L.Mangasarian et al., a modified SVM classification algorithm, two SVM regression (SVR) algorithms and a method for feature selection of minimal kernel classifiers are proposed respectively.Firstly, we introduce the mathematical theory of support vector machine (Statistical Learning Theory, SLT), the standard models, some algorithms for SVM and the development as well as some datasets, optimization softwares and some applications of SVM.In the second chapter, we discuss the SVM algorithms and propose a SOR-ASVM algorithm in order to improve the ASVM algorithm for solving the nonlinear- separable model.In the next chapter, an effective Newton-type SVR algorithm and a Lagrange SVR algorithm are presented by making use of the corresponding algorithms which were proposed for SVM.In the fourth chapter, the special structural SVM (minimal kernel classifier) is introduced and an improvement of the minimal kernel classifier is presented which perform feature selection automatically.In the last chapter, we summarize the paper's contents and propose some suggestions of the future work.