본문 바로가기

Understanding Black-box Predictions via Influence Functions1