Redundant Feature Elimination, a Supervised Solution

摘要：

It is a challenging problem to analyze high dimensional data sets like microarray data, where many irrelevant and weakly relevant but redundant features hurt generalization performance of classifiers. Irrelevant features are considered in most of previous works, are removed by considering label information. Weakly relevant but redundant features are only considered by a few works, which considered using linear or nonlinear filters. But these filters do not consider utilizing the label information to obtain discriminative contribution of each feature and furthermore eliminate those features with little discriminative contribution. Here we propose a novel supervised metric based on discriminative contribution to perform redundant feature elimination. By the new metric, complementary features are likely to be preserved, which is beneficial for the final classification. Experimental results on three microarray data sets show our proposed metric for redundant feature elimination based on discriminative contribution is better than the previous state-of-arts linear or nonlinear metrics on the problem of analysis of microarray data sets.

关键词： Machine Learning Data Mining Feature Selection Bioinformatics Microarray Analysis

作者: Guo-Zheng Li Xue-Qiang Zeng

作者单位: Department of Control Science & Engineering,Tongji University,Shanghai 201804,China;School of Comput School of Computer Engineering and Science,Shanghai University,Shanghai 200072,China

会议类型: 国际会议

会议名称: 2008高等智能国际会议(2008 International Conference on Advanced Intelligence)

会议地点: 北京

会议语种:英文

在线出版日期: 2008-10-18（万方平台首次上网日期，不代表论文的发表时间）