摘要
在支持向量机的学习过程中,有些情况下训练样本不能一次全部给出,这样当有新样本加入训练集时,支持向量集和训练样本集的等价关系将被打破。为了解决这个问题,本文提出了有新样本加入的支持向量机的学习策略。通过对新样本的分析,选出能代替原样本和新样本进行学习的样本,并给出这些样本应满足的条件,最后给出了相应的学习策略。对标准数据集的实验表明,本学习策略可以在新增样本增加后,有效压缩样本集的大小,提高分类的速度,舍弃无用的样本,同时保证了分类精度。
Sometimes an entire training sample can not be given at a time,in this case the equivalence between the support vector set and training set will be broken when new samples are introduced into the training set.In order to solve this problem this paper proposes a learning strategy of support vector machine introducing new samples.By analyzing the new samples,the samples replacing the old and new samples for learning are selected and the condition to satisfy these samples is given.Finally,the strategy of learning is given.The experimental results with the standard dataset show that the training time is greatly reduced while the classification precision is guaranteed.
出处
《河南科技大学学报（自然科学版）》
CAS
2007年第5期70-72,共3页
Journal of Henan University of Science And Technology:Natural Science
基金
国家自然科学基金项目(60574075)
关键词
支持向量机(SVM)
KKT条件
分类
新样本
Support vector machine
Karush-Kuhn-Tucker conditions
Classification
New samples
