Regularizing Black-box Models for Improved Interpretability (1902.06787v6)
Abstract: Most of the work on interpretable machine learning has focused on designing either inherently interpretable models, which typically trade-off accuracy for interpretability, or post-hoc explanation systems, whose explanation quality can be unpredictable. Our method, ExpO, is a hybridization of these approaches that regularizes a model for explanation quality at training time. Importantly, these regularizers are differentiable, model agnostic, and require no domain knowledge to define. We demonstrate that post-hoc explanations for ExpO-regularized models have better explanation quality, as measured by the common fidelity and stability metrics. We verify that improving these metrics leads to significantly more useful explanations with a user study on a realistic task.
- Gregory Plumb (11 papers)
- Maruan Al-Shedivat (20 papers)
- Adam Perer (29 papers)
- Eric Xing (127 papers)
- Ameet Talwalkar (89 papers)
- Angel Alexander Cabrera (1 paper)