Machine Learning model interpretability vs model performance… What’s the right balance? Here’s a great presentation on applying Bayesian inference to generate human-interpretable decision sets: Human in the Loop: Bayesian Rules to Enable Explainable AI (pdf)