Early Stopping for Mutual Information Based Feature Selection

A popular method for feature selection are filters based on the estimation of the mutual information between the features and the target. If the data is very high dimensional, even simple, iterative methods require substantial computational time. In this work we propose an early stopping method for feature selectors that reduces the complexity of the feature selector by orders of magnitute without any loss of predictive performance. We demonstrate the practical use of early stopping on high dimensional image clasification tasks.