Rethinking Data Shapley for Data Selection Tasks: Misleads and Merits

arXiv:2405.03875v1 Announce Type: new
Abstract: Data Shapley provides a principled approach to data valuation and plays a crucial role in data-centric machine learning (ML) research. Data selection is considered a standard application of Data Shapley. However, its data selection performance has shown to be inconsistent across settings in the literature. This study aims to deepen our understanding of this phenomenon. We introduce a hypothesis testing framework and show that Data Shapley's performance can be no better than random selection without …

abstract application arxiv cs.lg data data-centric however literature machine machine learning performance research role standard stat.ml tasks type valuation