Limits of Deepfake Detection: A Robust Estimation Viewpoint

TL;DR

This paper analyzes the fundamental limits of deepfake detection using robust statistics, providing bounds on error probabilities and linking them to epidemic thresholds in networks, thus offering a theoretical framework for understanding detection challenges.

Contribution

It introduces a robust statistical approach to bound deepfake detection error probabilities and connects these bounds to epidemic thresholds in network spreading processes.

Findings

Derived bounds on deepfake detection error probabilities.

Simplified bounds using Euclidean approximation for low error regimes.

Established relationships between detection errors and epidemic thresholds.

Abstract

Deepfake detection is formulated as a hypothesis testing problem to classify an image as genuine or GAN-generated. A robust statistics view of GANs is considered to bound the error probability for various GAN implementations in terms of their performance. The bounds are further simplified using a Euclidean approximation for the low error regime. Lastly, relationships between error probability and epidemic thresholds for spreading processes in networks are established.