A0980
Title: Faithful group Shapley value
Authors: Weijing Tang - Carnegie Mellon University (United States) [presenting]
Abstract: Data Shapley is an important tool for data valuation, which quantifies the contribution of individual data points to machine learning models. In practice, group-level data valuation is desirable when data providers contribute data in batches. However, it is identified that existing group-level extensions of data Shapley are vulnerable to shell company attacks, where strategic group splitting can unfairly inflate valuations. Faithful group Shapley value (FGSV) is proposed, which uniquely defends against such attacks. Building on original mathematical insights, a provably fast and accurate approximation algorithm is developed for computing FGSV. Empirical experiments demonstrate that the algorithm significantly outperforms state-of-the-art methods in computational efficiency and approximation accuracy, while ensuring faithful group-level valuation.
