MIKM: A mutual information-based K-medoids approach for feature selection

We propose a mutual information-based K-medoids approach (MIKM) for unsupervised and supervised feature selection, MIKM adopts mutual information (MI) to measure similarity between two features and applies K-medoids clustering to find representatives of feature clusters. The method partitions the original feature set into some distinct subsets or clusters such that the features within a cluster are highly similar to each other while those in different clusters are dissimilar, Each obtained representative is one of the original features. Consequently, The obtained representatives form a subset of the original features. Experimental results show that our proposed method can work more effectively than other methods.