Overview of Optimization Algorithms for Large-scale Support Vector Machines

Support vector machine (SVM) is one of the most classical machine learning algorithms, which performs well in many fields. However, the traditional training algorithms are not satisfactory in dealing with big data. Therefore, many algorithms for large-scale datasets have been proposed. Considering the classification problems, this paper introduces the main SVM training algorithms for large-scale datasets including the optimization methods of binary-class problems and multi-class problems.