Mutual information-based feature selection for multilabel classification
Mutual information-based feature selection for multilabel classification
复制标题
DOI：
10.1016/j.neucom.2013.06.035
复制
发表时间：
2013-12-25
期刊：
影响因子：
6
通讯作者：
Verleysen, Michel
中科院分区：
文献类型：
--
作者：
Doquire, Gauthier;Verleysen, Michel
本文介绍了一种新的方法来执行多标签分类问题的特征选择。与以前的作品的基础上的卡（2）统计量，所提出的方法使用的多元互信息标准结合的问题转换和修剪策略。这允许我们在特征选择过程中考虑类标签之间和特征之间可能的依赖关系。提出了一种基于排列测试结合重排序策略的剪枝参数自动设置方法。在人工和真实世界的数据集上进行的实验表明，我们的方法比现有的方法的兴趣。(C)2013爱思唯尔有限公司版权所有。
This paper introduces a new methodology to perform feature selection in multi-label classification problems. Unlike previous works based on the chi(2) statistics, the proposed approach uses the multivariate mutual information criterion combined with a problem transformation and a pruning strategy. This allows us to consider the possible dependencies between the class labels and between the features during the feature selection process. A way to automatically set the pruning parameter is also proposed, based on the permutation test combined with a resampling strategy. Experiments carried out on both artificial and real-world datasets show the interest of our approach over existing methods. (C) 2013 Elsevier B.V. All rights reserved.