Towards Quantitative Evaluation of Explainable AI Methods for Deepfake Detection

arXiv:2404.18649v1 Announce Type: new
Abstract: In this paper we propose a new framework for evaluating the performance of explanation methods on the decisions of a deepfake detector. This framework assesses the ability of an explanation method to spot the regions of a fake image with the biggest influence on the decision of the deepfake detector, by examining the extent to which these regions can be modified through a set of adversarial attacks, in order to flip the detector's prediction or …

abstract arxiv cs.ai cs.cv decisions deepfake detection evaluation explainable ai fake framework image influence paper performance quantitative spot type