Theory Of Support Vector Machine And Its Application

The support vector machine in the classification problem has a very good application, this paper mainly studies the theory of support vector machine. In this thesis, we will discuss the replaced the hinge loss functions of the support vector machine, and consider the nuclear logistic regression model. We show that, nuclear logistic regression model in the implementation of two class classification support vector machine function. In addition, also using nuclear logistic regression model provides a basis for probability estimation. In the nucleus of a logistic regression model based on, we propose a new classification method, called the import vector machine. And support vector machine is similar to it, using only a fraction of the training data and the kernel function, support vector machine and it is the amount of data required to be smaller and more. This makes the import vector machine in the data set in certain circumstances, significantly better than the support vector machine.