基于分层并行筛选样本的SVM增量学习算法  

An Improved Incremental Learning Algorithm Based on Hierarchical Filtering

作　　者：姜雪[1] 陶亮[1] 王华彬[1] 武杰[1] 

机构地区：[1]安徽大学计算机科学与技术学院,安徽合肥230039

出　　处：《计算机技术与发展》2007年第11期92-95,102,共5页Computer Technology and Development

基　　金：国家自然科学基金(60572128);安徽省人才开发基金(2005Z029);安徽大学人才队伍建设项目和创新团队基金

摘　　要：在增量学习过程中,随着训练集规模的增大,支持向量机的学习过程需要占用大量内存,寻优速度非常缓慢。在现有的一种支持向量机增量学习算法的基础上,结合并行学习思想,提出了一种分层并行筛选训练样本的支持向量机增量学习算法。理论分析和实验结果表明:与原有的算法相比,新算法能在保证支持向量机的分类能力的前提下显著提高训练速度。During the incremental learning, with the increase of the training set, it is very costly to process these data in terms of time and memory consumption. Based on the existing incremental learning algorithm for support vector machine（SVM）and joined with the idea of the parallel learning, a novel algorithm of incremental learning is proposed, which filters the training samples in a hierarchical and parallel way. The theoretical analysis and experiment results show that, compared with the original one, the new algorithm is able to improve the speed of SVM greatly, while the ability of SVM to classification is guaranteed.

关 键 词：支持向量机 机器学习 分类 增量学习 分层筛选 

分 类 号：TP181[自动化与计算机技术—控制理论与控制工程]