Keywords: interpretability, shapley, amortization, explainability, game theory
Abstract: Although Shapley values are theoretically appealing for explaining black-box models, they are costly to calculate and thus impractical in settings that involve large, high-dimensional models. To remedy this issue, we introduce FastSHAP, a new method for estimating Shapley values in a single forward pass using a learned explainer model. To enable efficient training without requiring ground truth Shapley values, we develop an approach to train FastSHAP via stochastic gradient descent using a weighted least-squares objective function. In our experiments with tabular and image datasets, we compare FastSHAP to existing estimation approaches and find that it generates accurate explanations with an orders-of-magnitude speedup.
One-sentence Summary: We introduce FastSHAP, a new method for estimating Shapley values in a single forward pass using an explainer model that is learned via stochastic gradient optimization using a weighted least squares-like objective function.
Supplementary Material: zip