PDD-SHAP: Fast Approximations for Shapley Values using Functional Decomposition
This addresses the problem of high computational cost for practitioners needing to explain many predictions from black-box models, representing an incremental improvement over existing methods.
The paper tackles the computational expense of computing Shapley values for explaining black-box model predictions by proposing PDD-SHAP, an algorithm that approximates the model using functional decomposition, resulting in orders of magnitude faster calculations for large datasets.
Because of their strong theoretical properties, Shapley values have become very popular as a way to explain predictions made by black box models. Unfortuately, most existing techniques to compute Shapley values are computationally very expensive. We propose PDD-SHAP, an algorithm that uses an ANOVA-based functional decomposition model to approximate the black-box model being explained. This allows us to calculate Shapley values orders of magnitude faster than existing methods for large datasets, significantly reducing the amortized cost of computing Shapley values when many predictions need to be explained.