支持向量机的凸优化求解

Convex optimization of support vector machines

支持向量机(SVM)是一种基于统计学习理论的机器学习方法,由于其优越的学习性能,已经成为当前模式识别、数据挖掘、大数据处理等机器学习领域的研究热点.查阅相关同类文章,发现其中对SVM理论中公式,如距离函数d、拉格朗日函数L(w,b,α)、二次凸优化函数f(x)等的来龙去脉缺少细致的阐述.本文对SVM理论中典型的线性最优二分类问题的求解进行了完整的推导,并给出了对岩屑岩性分类识别的结果,也为今后的非线性多类模式分解作出铺垫.

Support Vector Machine （SVM） is a machine learning method based on statistical learning the- ory. Because of its superior learning performance. It has become the hot topic of pattern recognition, data mining, machine learning and other large data processing areas. Access to relevant similar articles. Found that the theory of SVM formula,such as the distance function d , Lagrange function L（w,b,a） , quadratic convex optimization function f（x） lack detailed exposition. This paper tries to make readers to the typical linear optimal binary classification problem have a complete idea. The results of the classifica- tion and identification of rock cuttings are presented. It also wants to make a foundation of nonlinear multi-class pattern decomposition.