基于提升的BAN组合分类器

提升是一种有效的分类器组合方法,它能够提高不稳定学习算法的分类性能,但对稳定的学习算法效果不明显.BAN(BN augmented Nave-Bayes)是一种增强的贝叶斯网络分类器,通过提升很容易提高其分类性能.文中比较了GBN(general BN)和BAN的打包分类器Wrapping-BAN-GBN与基于提升的BAN组合分类器Boosting-BAN,最后通过实验结果显示了在大多数实验数据上,Boosting-BAN分类器显示出较高的分类正确率.

Boosting-based BAN Combination Classifier

Boosting is an effective classifier combination method, which can improve classification performance of an unstable learning algorithm. But it dose not make much more improvement on a stable learning algorithm. BAN, i.e. BN augmented Naive-Bayes, is an augmented Bayesian network classifier, whose accuracy is easy to improve by the Boosting technique. In this paper, a wrapping classifier which wraps around GBN and BAN is compared with the Boosting-BAN classifier which is Boosting based BAN combination classifier. Finally, experimental results show that the Boostlng-BAN has higher classification accuracy on most data sets.