Computing Conditional Shapley Values Using Tabular Foundation Models

TL;DR

This paper introduces a novel approach to compute conditional Shapley values efficiently using tabular foundation models like TabPFN, significantly reducing computation time while maintaining high accuracy compared to existing methods.

Contribution

The paper demonstrates how tabular foundation models can be used to estimate conditional Shapley values without retraining, outperforming traditional methods in speed and often in accuracy.

Findings

TabPFN-based methods often outperform state-of-the-art in accuracy.

TabPFN achieves comparable results with much lower runtime.

The approach is effective on both simulated and real datasets.

Abstract

Shapley values have become a cornerstone of explainable AI, but they are computationally expensive to use, especially when features are dependent. Evaluating them requires approximating a large number of conditional expectations, either via Monte Carlo integration or regression. Until recently it has not been possible to fully exploit deep learning for the regression approach, because retraining for each conditional expectation takes too long. Tabular foundation models such as TabPFN overcome this computational hurdle by leveraging in-context learning, so each conditional expectation can be approximated without any re-training. In this paper, we compute Shapley values with multiple variants of TabPFN and compare their performance with state-of-the-art methods on both simulated and real datasets. In most cases, TabPFN yields the best performance; where it does not, it is only marginally worse than the best method, at a fraction of the runtime. We discuss further improvements and how tabular foundation models can be better adapted specifically for conditional Shapley value estimation.