The Study Of Multi-field Adaptive Classfication Based On Support Vector Machine

Support vector machine is a good learning machine based on the statistical learning theory.In linear classification, it find the super hyperplane between with two-class samples in the input space. In nonlinear classification, it map the input data into a high-dimensional feature space, and then find the super hyperplane in feature space. A good property of SVM is that we need not compute the mapped patterns explicitly, and instead we only need the dot products between mapped patters, which are directly available from the kernel function. It combine hyperplane of maximizing margin, Mercer kernel with flabby variable and the sparse solution, so have good performance in many defiant problem about machine learning. In order to gain the best generalization ability, SVM find the best compromise between with complexity of model and learning ability by using the information of finite samples.The Kernel function is an important part of support vector machine, and it is key problem of SVM research. Different kernel function can produce different SVM. Based on this, the kernel function of SVM is studied in this thesis. The main research works of this thesis are listed as below:First, this paper introduces VC dimension theory and structural risk minimization principle. Introduce kernel function through SVM and discuss the importance of selecting parameters in kernel function. As Gauss kernel function for an example, we discuss the influences of Gauss kernel radius and punishment-parameter C in SVM.Second, there are two kinds of kernel function: local kernel function and global kernel function. We propose a novel adaptive compound kernel function by taking advantage of local kernel function and global kernel function.Next, the adaptive compound kernel function is applied in SVM and compared with other kernels in different fields experiments. The experimental results show that the novel kernel can achieve better performance than other kernels.Last, on the basis of experiment data, we build a multi-filed adaptive model library based on SVM.