摘要
在如何从海量的数据中提取有用的信息上提出了一种新的SVM的增量学习算法。该算法基于KKT条件,通过研究支持向量分布特点,分析了新样本加入训练集后,支持向量集的变化情况,提出等势训练集的观点。能对训练数据进行有效的遗忘淘汰,使得学习对象的知识得到了积累。在理论分析和对旅游信息分类的应用结果表明,该算法能在保持分类精度的同时,有效得提高训练速度。
A new incremental learning algorithm of support vector machine is proposed to extract useful information from mass data. Based on KKT condition, the possible change of support vector set is analyed after new samples are added to training set and a view named parallel potential data set is put forward. Distributed knowledge of the training samples is accumulated and samples is discarded optimally. The theoretical analysis and experimental results to traveling information classification shows that this algorithm improve training speed when maintaining the precision.
出处
《计算机工程与设计》
CSCD
北大核心
2007年第3期700-702,共3页
Computer Engineering and Design
基金
国家自然科学基金项目(60442003)
北京市自然科学基金项目(4042012)
北京市教育委员会科技发展计划基金项目(KZ200510011009)
关键词
支持向量机
增量学习
旅游信息
分类
KKT条件
support vector machine
incremental learning
traveling information
classification
KKT condition
