ScalableRule-Based RepresentationLearningfor InterpretableClassification