An improved TANC method based on Bayesian equivalence theorem | |
Zhao, Xiaoq-Iang; Yang, Jia-Min; Zhou, Jin-Hu | |
刊名 | Information Technology Journal |
2013 | |
卷号 | 12期号:17页码:4336-4339 |
关键词 | Classifiers Bayes Classifier Classification accuracy Dependency relationship Equivalence theorem Generalization ability High-efficiency TANC Tree augmented Naive Bayes classifiers |
ISSN号 | 18125638 |
DOI | 10.3923/itj.2013.4336.4339 |
英文摘要 | TANC (Tree Augmented Naive Bayes Classifier) is efficient extension of NBC (Naive Bayes Classifier). This method not only inherits the simple and high efficiency performances of NBC, but also enhances the generalization ability. However it ignores the correlation between weaken attributes. So, an improved TANC method is proposed in this study according the dependence degree and the correlation between attributes. This method can set up a correspond dependencies to effectively improve the classification accuracy by selecting the appropriate attributes. Compared with NBC and TANC, experimental results showed this method is better than TANC and NBC in performance. © 2013 Asian Network for Scientific Information. |
语种 | 英语 |
出版者 | Asian Network for Scientific Information, 308-Lasani Town, Sargodha Road, Faisalabad, Pakistan |
内容类型 | 期刊论文 |
源URL | [http://ir.lut.edu.cn/handle/2XXMBERH/113312] |
专题 | 兰州理工大学 |
作者单位 | College of Electrical Engineering an Information Engineering, Lanzhou University of Technology, Lanzhou, 730050, Gansu, China |
推荐引用方式 GB/T 7714 | Zhao, Xiaoq-Iang,Yang, Jia-Min,Zhou, Jin-Hu. An improved TANC method based on Bayesian equivalence theorem[J]. Information Technology Journal,2013,12(17):4336-4339. |
APA | Zhao, Xiaoq-Iang,Yang, Jia-Min,&Zhou, Jin-Hu.(2013).An improved TANC method based on Bayesian equivalence theorem.Information Technology Journal,12(17),4336-4339. |
MLA | Zhao, Xiaoq-Iang,et al."An improved TANC method based on Bayesian equivalence theorem".Information Technology Journal 12.17(2013):4336-4339. |
个性服务 |
查看访问统计 |
相关权益政策 |
暂无数据 |
收藏/分享 |
除非特别说明,本系统中所有内容都受版权保护,并保留所有权利。
修改评论