rule-based-models

This paper introduces CDPR, a new approach for learning highly accurate and interpretable rule sets for classification using submodular maximization, achieving over 2.5x improvement in coverage compared to existing methods.