What does this research mean for the field?

Question

Accepted Answer

Post-hoc explanations from popular XAI methods, including SHAP, LIME, and DiCE, leak sensitive data and enable feasible membership inference and training data extraction attacks. Novelty: ClaimNovelty.INCREMENTAL. Consensus alignment: ConsensusAlignment.NEUTRAL.

On the antagonism of explainability and privacy: A comparative study of attacks and explainers

Discussion

Member takes

Overview

Key Points

Cite This Study