| Support Vector Machine ( for short SVM ) is a new method of Machine Learning, it has became one research focus in the field of international Machine Learning because of its excellent learning capacity. Support Vector Regression ( for short SVR ) is an important branch of SVM, SVR has been applied to system identification, nonlinear system prediction and good results have been demonstrated.In this paper, with image representation by support vector regression (SVR) ,we make an research on image process of SVR image. Making use of the characters of SVR image and all kinds of traditional edge detection methods, the methods of edge detection based SVR image have been proposed, which differ from operator methods and are easier to understand and realize. Satisfactory results were obtained.By combining the traditional neural network method with the SVR theory, this paper also addresses the problem of boundary value problems and replaces the neural network function with SVR regression function. Some examples are given to illustrate the actual application process, the analysis of which demonstrates SVR's validity in solving boundary value problems. |
