摘要
针对兼类样本,提出一种类增量学习算法.利用超球支持向量机,对每类样本求得一个能包围该类尽可能多样本的最小超球,使各类样本之间通过超球隔开.增量学习时,对新增样本以及旧样本集中的支持向量和超球附近的非支持向量进行训练,使得算法在很小的空间代价下实现兼类样本类增量学习.分类过程中,根据待分类样本到各超球球心的距离判定其所属类别.实验结果表明,该算法具有较快的训练、分类速度和较高的分类精度.
To multi-class sample, an incremental learning algorithm is proposed in this paper. Hyper-sphere support vector machine is used to get the smallest hyper-sphere that contains most samples of a class, which can divide the class samples from others. In the process of class incremental learning, the new samples, the history support vectors and the history samples that near the hyper-sphere are trained. Therefore, the multi-class incremental learning can be realized in a small memory space. For the sample to be classified, the distances from it to the centre of every hypersphere are used to confirm the classes that the sample belongs to. The experimental results show that the algorithm has a higher performance on training speed, classification speed, and classification precision.
出处
《控制与决策》
EI
CSCD
北大核心
2009年第1期137-140,共4页
Control and Decision
基金
国家自然科学基金项目(60603023)
国家973计划项目(2001CCA00700)
关键词
支持向量机
超球
兼类
类增量学习
Support vector machines
Hyper-sphere
Multi-class
Class incremental learning
