支持向量机分类理论中的几个细节理解及实验
Understanding of Theory and Experiments in Support Vector Machine Classification

目前的大数据时代，机器学习作为数据处理的关键技术，不可缺少。各个高校在很多专业都在开展机器学习这门课程，其中的支持向量机算法作为重要内容之一，广受学生青睐。但在教学过程中，我们发现其理论在某些方面比较晦涩难懂，本文在简要概述完支持向量机分类的理论后，总结了几个关键细节的理解，比如：支持向量机线性分类器模型中的最大化Margin的理解；线性分类器模型中目标函数和约束条件的推导过程中某些知识点的理解；线性分类器模型求解过程中，为什么拉格朗日乘子大于0对应的数据点就是支持向量？支持向量机的分类面唯一吗？最后，给出了Matlab、R语言、python等几种软件在iris数据上的实验代码及实验中得到的支持向量机模型的获取方法。这些细节的理解及软件实验，对支持向量机模型都进行了透彻的剖析和融会贯通，有助于支持向量机分类学习中的初步探索和深入的研究，给广大支持向量机的学习者和使用者提供了重要的参考。
In the era of big data, machine learning is indispensable for data processing. Various colleges and universities are carrying out the course of machine learning in many majors. As one of the im-portant contents, support vector machine algorithm is widely favored by students. But in the pro-cess of teaching, we find that its theory is rather obscure in some aspects. After summarizing the theory of support vector machine classification, this paper summarizes several key points’ under-standing, such as, the understanding of maximizing margin in support vector machine linear classi-fier model; the understanding of some knowledge points in the derivation of objective functions and constraints in linear classifier model; in the process of solving linear classifier model, why is the da-ta point corresponding to Lagrange multiplier greater than 0 support vector? Is the support vector machine classification face unique? Then, the experimental codes of several softwares such as MATLAB, R language and python on iris data and the understanding of relevant results are ex-pounded. The understanding of these details and the software experiment have carried on the thorough analysis and the fusion to the support vector machine model, which is helpful for the pre-liminary exploration and in-depth study of SVM classification learning, having provided the im-portant reference for the broad support vector machine learner and the user.