1、 湖南大学毕业设计（论文） 第 I 页支持向量机实验模型研究与设计摘 要支持向量机作为一种优秀的学习方法，它具有理论完备、适应性强、全局优化、训练时间短、泛化性能好等优点，已成为当前国际机器学习界的研究热点，有着很好的应用前景。支持向量机方法与机器学习，神经网络等学科现有的理论和方法相比，有明显的优越性，特别是它局部最优解一定是全局最优解，克服了限于局部最小问题，这是人工神经网络学习算法所不能及的。本文论述了支持向量机的基本原理和思想，侧重分析了支持向量机的 SMO 算法（即序列最小优化（SMO ）算法）和三种核函数（即线性内核、径向基函数内核、多项式内核） 。本课题研究并实现了这种算法，并使
2、用这种训练算法和这三种核函数实现了基于支持向量机的机器学习模型，实现了使用这种算法和三个核函数中的一个对数据集进行分类。此模型能够对线性可分的和非线性可分的两种情况的两类数据进行正确分类，并对分类的效果进行分析。此模型不仅能够装载已有的数据进行分类，还实现了手动创建数据集的功能，可以对手动创建的数据集进行分类。关键词：支持向量机，分类，训练算法，核函数湖南大学毕业设计（论文） 第 II 页The Study and Design of Experiment Modle Based on Support Vector MachineAuthor: Su HanmuTutor: Zhong Qin
3、gliuAbstractSupport Vector Machine as a kind of outstanding Machine Learning methods, has many advantages such as a complete theory ,a good compatibility ,a overall optimization, a short training time, a good generalized capability and so on. So that, it already become the research focus in the fiel
4、d of international Machine Learning. The application prospect of Support Vector Machine is very good. Compared with other theory and method of Machine Learning Support Vector Machine has obvious superiority. Especially its partial optimal answer is the overall optimal answer. This is what artificial
5、 neural network can not solve.This text has expounded the basic principle and thought of Support Vector Machine, and has laid particular emphasi on analyzing one kind training algorithms (Sequential Minimal optimizer algorithm) and three kernel functions(Linear, Polynomial, RBF) of Support Vector Ma
6、chine. This thesis research and implement the three training algorithms, and then, implement the Machine Learning model based on Support Vector Machine. You can choose one training algorithms and one of the three kernel functions to classify the data correctly by the model. The model can classify tw
7、o data sets no matter whether the sets is linear or non-linear, and analyse the result of classification. The model can classify the data sets which is loaded from the existent one. The model implement the function of creating data sets by user, and then classify the data sets.Key words: Support Vector Machine, classify, training algorithm, kernel function
