序贯最小优化的改进算法  

An Improvement Algorithm to Sequential Minimal Optimization

作　　者：李建民[1,2] 张钹[1,2] 林福宗[1,2] 

机构地区：[1]清华大学计算机科学与技术系,北京100084 [2]清华大学智能技术与系统国家重点实验室,北京100084

出　　处：《软件学报》2003年第5期918-924,共7页Journal of Software

基　　金：国家自然科学基金;国家重点基础研究发展规划(973)~~

摘　　要：序贯最小优化(sequential minimal optimization,简称SMO)算法是目前解决大量数据下支持向量机(support vector machine,简称SVM)训练问题的一种十分有效的方法,但是确定工作集的可行方向策略会降低缓存的效率.给出了SMO的一种可行方向法的解释,进而提出了一种收益代价平衡的工作集选择方法,综合考虑与工作集相关的目标函数的下降量和计算代价,以提高缓存的效率.实验结果表明,该方法可以提高SMO算法的性能,缩短SVM分类器的训练时间,特别适用于样本较多、支持向量较多、非有界支持向量较多的情况.At present sequential minimal optimization (SMO) algorithm is a quite efficient method for training large-scale support vector machines (SVM). However, the feasible direction strategy for selecting working sets may degrade the performance of the kernel cache maintained in SMO. After an interpretation of SMO as the feasible direction method in the traditional optimization theory, a novel strategy for selecting working sets applied in SMO is presented. Based on the original feasible direction selection strategy, the new method takes both reduction of the object function and computational cost related to the selected working set into consideration in order to improve the efficiency of the kernel cache. It is shown in the experiments on the well-known data sets that computation of the kernel function and training time is reduced greatly, especially for the problems with many samples, support vectors and non-bound support vectors.

关 键 词：支持向量机 机器学习 序贯最小优化 SMO算法 

分 类 号：TP181[自动化与计算机技术—控制理论与控制工程]