Limits of Deepfake Detection: A Robust Estimation Viewpoint
This work addresses the challenge of detecting AI-generated images for security and media integrity applications, but appears incremental as it builds on existing statistical frameworks.
The paper tackles the problem of deepfake detection by formulating it as a hypothesis testing problem and using robust statistics to bound error probabilities for various GAN implementations, with results including simplified bounds using Euclidean approximations and connections to epidemic thresholds in networks.
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.