统一框架下软计算方法和支撑矢量机的机理研究

MECHANISM STUDY ON SOFT COMPUTING PARADIGMS AND SUPPORT VECTOR MACHINE UNDER A UNIFIED FRAMEWORK

本文在加权径向基函数这一统一方法的框架下,对常用的几种软计算方法(包括神经网络、小波网络、模糊系统、贝叶斯分类器、模糊划分)以及支撑矢量机等机器学习方法的内在机理做了进一步的研究.特别地,对于支撑矢量机这种新的学习方法,文中分析了它和神经网络等方法之间的异同处,并尝试性地把其纳入统一的框架下,从而为支撑矢量机及软计算方法的研究与应用提供了理论上的指导.

Under a unified frame, this paper deals with the mechanism study on supprot vector machine and several existing soft computing paradigms, including neural and wavelets networks, fuzzy systems, Bayesian classifiers, and fuzzy partition, and on support vector machine, which is a novel learning technique based on statistical learning theory. Particularly the similarities and differences between suppert vector machine and the existing paradigms are analyzed, and they are tried to be brought into a unified framework. It is hoped that this paper would provide theoretical guide for studying and applying support vector machine and soft computing paradigms.