基于样本取样的SMO算法
A Sampling-based SMO Algorithm
-
摘要: 介绍了一种对样本集取样的方法，并在此基础上对序贯最小优化（sequential minimal optimization，SMO）算法进行了改进，提出了取样序贯最小优化（SSMO）算法．SSMO算法去掉了大部分非支持向量，将支持向量逐渐收集到工作集中．实验结果表明，该方法提高了SMO算法的性能，缩短了支持向量机分类器的训练时间．Abstract: An algorithm of sampling data set is introduced.Se quential minimal optimization(SMO)algorithm is improved based on the sampling algorithm.Then a sampling-based SMO(S-SMO)algorithm is presented.The S-SMO algorithm reduces a majority of non-support vectors and support vectors are collected into work set gradually.It is shown in the experiments that S-SMO algorithm improves the performance of SMO and the time of training SVM classifier is reduced greatly.
下载:
