The Construction Of Vector-valued Reproducing Kernel Space And Its Application In Network Intrusion Detection

Kernel method is a very effective method for solving nonlinear model analysis. It is also a hot research topic in machine learning field. Kernel function is the key factor that affects the performance of nuclear method, and the core function is widely used in practical problems,such as: industry, finance, military, transportation, aerospace, etc. At present, more and more scholars begin to engage in the study of kernel machine learning and show a strong interest in it. But it is difficult to conduct a liner machine learning, because most of them are nonlinear in practice. Therefore, it will be of great practical significance to create a new method that transform nonlinear problem into linear.In this paper, we constructed the kernel function of vector valued reproducing kernel,and it were applied to the network intrusion detection by using the support vector machine algorithm.First of all, we introduce the basic properties and theorems of the reproducing kernel space, and give the concrete expression of the reproducing kernel function in the reproducing kernel space. By using the property of the scalar value of reproducing kernel, we establish the Hilbert space of the vector valued reproducing kernel. The ability of the support vector machine classification is mainly affected by the kernel function and the error penalty parameter for the given training data. Thus, we construct the vector valued reproducing kernel Hilbert space which is available for data analysis. At last, vector valued reproducing kernel functions will be applied in the support vector machine algorithm. On the one hand, from three aspects: the construction of the vector valued kernel space, the construction of the kernel function, and the parametric selection of the vector valued kernel functions, they could illustrate that different kernel functions affect the amount of support vectors combine with the support vector machine algorithm. On the other hand, we apply a multi-core vector valued kernel function that given in this paper to the network intrusion detection by using the support vector machine algorithm.