Abstract:
This paper modifies the query-by-committee(QBC) method of active learning by combining vote entropy and kullback-leibler divergence for learning TAN classifier to model telecom clients’ credit classification. Experimental results show that the proposed QBC is better than previous QBC approach in classification accuracy with a few labeled training examples.
Key words:
Active learning,
Query-by-committee(QBC),
Vote entropy,
Kullback-leibler divergence,
Credit classification
摘要: 结合委员会成员投票熵和相对熵,改进了基于委员会选择算法(QBC)的主动学习,并应用基于该算法的主动贝叶斯网络对电信客户信用风险分类进行建模。实验结果表明,提出的基于改进的QBC主动贝叶斯网络分类器所建模型比原有算法有更好的分类精度,并且使用了少量的训练数据。
关键词:
主动学习,
委员会选择,
投票熵,
相对熵,
信用分类
ZHAO Yue; MU Zhichun. Research of Query-by-committee Method of Active Learning[J]. Computer Engineering, 2006, 32(24): 23-25.
赵 悦;穆志纯. 基于QBC的主动学习研究及其应用[J]. 计算机工程, 2006, 32(24): 23-25.