Can High-Order Dependencies Improve Mutual Information based Feature Selection?

Mutual information (MI) based approaches are a popular paradigm for feature selection. Most previous methods have made use of low-dimensional MI quantities that are only effective at detecting low-order dependencies be-tween variables. Several works have considered the use of higher dimensional mutual information, but the theoretical underpinning of these approaches is not yet comprehensive. To fill this gap, in this paper, we systematically investigate the issues of employing high-order dependencies for mutual infor-mation based feature selection. We first identify a set of assumptions under which the original high-dimensional mutual information based criterion can be decomposed into a set of low-dimensional MI quantities. By relaxing these assumptions, we arrive at a principled approach for constructing higher di-mensional MI based feature selection methods that takes into account higher order feature interactions. Our extensive experimental evaluation on real data sets provides concrete evidence that methodological inclusion of high-order dependencies improve MI based feature selection