摘要
在应用基本的支持向量机算法的基础上,提出了一种新的分布增量学习方法,利用主动学习策略对训练样本进行选择,逐步增大提交给学习器训练样本的规模,以提高学习器的识别精确率。实验表明,采用主动学习策略的支持向量机算法是有效的,中文娱乐新闻词语识别的正确率和召回率分别达到了78.92%和86.42%,收到了良好的效果。
Based on applying the fundamental SVM algorithm,a new approach for distributed incremental learning is introduced in this paper,which adopts active learning strategy to select training sample and gradually increase the scale of training sample submitted to the learner,so that the recognition accuracy of the learner is improved.Experiment shows that the SVM algorithm using active learning strategy is effective,and the precision and recall rate of recognizing the words and expressions of Chinese entertainment news achieve up to 78.92% and 86.42% respectively.It has gained good effect.
出处
《计算机应用与软件》
CSCD
2011年第2期249-252,共4页
Computer Applications and Software
基金
忻州师范学院科研基金项目(200904)
关键词
机器学习
娱乐新闻词语
支持向量机
文本分类
主动学习
Machine learning Words and expressions of entertainment news Support vector machine（SVM） Text classification Active learning
