Support Vector Machines
Support Vector Machines
复制标题
DOI：
10.1201/9781003139041-11
复制
发表时间：
2021-03
期刊：
影响因子：
--
通讯作者：
Harry G. Perros
中科院分区：
文献类型：
--
作者：
Harry G. Perros
在这份报告中，我们介绍了支持向量机及其实现的概述。支持向量机（SVM）是一种有效的二进制分类机制，具有良好的泛化性能。支持向量机在高维特征空间中找到最佳线性分离器，即最大边缘超平面-这与神经网络等在原始空间中构建高度非线性分离器的机制不同。这项工作的目的是让作者对CS281 A中没有直接涉及的机器学习工具进行广泛的概述，CS281 A是加州大学伯克利分校概率图模型的研究生课程。其中包括计算学习理论、非参数方法、支持向量机、优化以及这些概念的实际实现等主题。
In this report we present an overview of Support Vector Machines and their implementations. Support Vector Machines (SVMs) are an effective mechanism for binary classification that have good generalization properties. SVMs find the optimal linear separator, the maximal margin hyperplane, in a high dimensional feature space – a different approach from mechanisms like Neural Networks that build a highly non-linear separator in the original space. The objective of this work was to give the author broad overview of Machine Learning tools not directly covered in CS281A,– a graduate course in probabilistic graphical models at UC Berkeley. This included topics such as computational learning theory, non-parametric methods, support vector machines, optimization and, finally, practical implementations of these concepts .