高维支持向量机的一些新发展

Some new developments of support vector machine in high dimension

对高维支持向量机(SVM)的一些新发展如非凸惩罚SVM,L_(1)范数SVM的误差界以及SVM在充分性降维中的应用进行了介绍;通过数值模拟和实例分析,展示了这些新方法在有限样本时的表现;讨论了一些可能的方向和问题.

Some selective new developments of support vector machine(SVM),including non-convex penalized SVM,the error bound of L_(1) norm SVM,and the application of SVM in sufficient dimension reduction are reviewed.The performance of these new methods in high-dimensional SVM is demonstrated by numerical simulation and real data analysis.Several possible new directions and issues are discussed.