Effective Feature Selection for Feature Possessing Group Structure

Abstract

Feature selection has become an interesting research topic in recent years. It is an effective method to tackle the data with high dimension. The underlying structure has been ignored by the previous feature selection method and it determines the feature individually. Considering this we focus on the problem where feature possess some group structure. To solve this problem we present group feature selection method at group level to execute feature selection. Its objective is to execute the feature selection in within the group and between the group of features that select discriminative features and remove redundant features to obtain optimal subset. We demonstrate our method on data sets and perform the task to achieve classification accuracy.