D4: Detection of Adversarial Diffusion Deepfakes Using Disjoint Ensembles
Preprint
Abstract
Detecting diffusion-generated deepfake images remains an open problem. Current detection methods fail against an adversary who adds imperceptible adversarial perturbations to the deepfake to evade detection. In this work, we propose Disjoint Diffusion Deepfake Detection (D4), a deepfake detector designed to improve black-box adversarial robustness beyond de facto solutions such as adversarial training. D4 uses an ensemble of models over disjoint subsets of the frequency spectrum to significantly improve adversarial robustness. Our key insight is to leverage a redundancy in the frequency domain and apply a saliency partitioning technique to disjointly distribute frequency components across multiple models. We formally prove that these disjoint ensembles lead to a reduction in the dimensionality of the input subspace where adversarial deepfakes lie, thereby making adversarial deepfakes harder to find for black-box attacks. We then empirically validate the D4 method against several black-box attacks and find that D4 significantly outperforms existing state-of-the-art defenses applied to diffusion-generated deepfake detection. We also demonstrate that D4 provides robustness against adversarial deepfakes from unseen data distributions as well as unseen generative techniques.Keywords
All Related Versions
This publication has 9 references indexed in Scilit:
- Towards the Detection of Diffusion Model Deepfakes2024
- DE-FAKE: Detection and Attribution of Fake Images Generated by Text-to-Image Generation Models2023
- GRAPHITE: Generating Automatic Physical Examples for Machine-Learning Attacks on Computer Vision Systems2022
- Exposing Vulnerabilities of Deepfake Detection Systems with Robust AttacksDigital Threats: Research and Practice, 2022
- EnsembleDet: ensembling against adversarial attack on deepfake detectionJournal of Electronic Imaging, 2021
- Imperceptible Adversarial Examples For Fake Image Detection2021
- Detecting and Simulating Artifacts in GAN Fake Images2019
- A Discussion of 'Adversarial Examples Are Not Bugs, They Are Features'Distill, 2019
- Recent Progress on Generative Adversarial Networks (GANs): A SurveyIEEE Access, 2019