Research on structure of support vector machine kernel function

Support vector machine is a newly developed learning machine based on statistical learning theory in recent years. The choice of kernel function is undoubtedly the core issue of utmost importance. At first the definition of kernel function was given,and then the nature of kernel theory was analyzed,finally the optimal choice of kernel function and its structure was clarified,which provides a theoretical reference on the choice of kernel function. This learning machine uses structural risk instead of experience risk thus owns an uncomparable advantage over the traditional neural network.