Towards A Rigorous Science of Interpretable Machine Learning