Research map / Label-efficient and robust learning / Self-supervised and contrastive
Self-Supervised Learning of Adversarial Example: Towards Good Generalizations for Deepfake Detection
Check your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| Use adversarially synthesized forgeries and configuration prediction | Xception deepfake detector · CelebDF | AUC: 0.734 -> 0.797 | Helped |
| Add adversarial augmentation without self-supervised tasks (ablation) | Xception · FF++ cross-method generalization | AUC (Avg.): 0.669 -> 0.709 | Helped |
| Replace adversarial augmentation with random configuration selection (ablation) | SLADD · FF++ cross-method generalization | AUC (Avg.): 0.769 -> 0.716 | Hurt |
| Use adversarial rather than random configuration selection (ablation) | SLADD · FF++ cross-method generalization | AUC (Avg.): 0.716 -> 0.769 | Helped |
| Replace forgery-specific synthesis with random general image operations (ablation) | SLADD · FF++ cross-method generalization | AUC (Avg.): 0.769 -> 0.710 | Hurt |
| Replace forgery-specific synthesis with synthesizer-selected general image operations (ablation) | SLADD · FF++ cross-method generalization | AUC (Avg.): 0.769 -> 0.711 | Hurt |
| Replace the proposed augmentation with fixed-region forgery augmentation (ablation) | SLADD · FF++ cross-method generalization | AUC (Avg.): 0.769 -> 0.717 | Hurt |
| Replace the proposed augmentation with another fixed-region forgery augmentation (ablation) | SLADD · FF++ cross-method generalization | AUC (Avg.): 0.769 -> 0.722 | Hurt |
| Add blending-ratio estimation as an auxiliary task (ablation) | SLADD · FF++ cross-method generalization | AUC (Avg.): 0.733 -> 0.769 | Helped |
| Add blending-type estimation as an auxiliary task (ablation) | SLADD · FF++ cross-method generalization | AUC (Avg.): 0.737 -> 0.769 | Helped |
| Add forgery-region estimation as an auxiliary task (ablation) | SLADD · FF++ cross-method generalization | AUC (Avg.): 0.724 -> 0.769 | Helped |
| Add all three self-supervised auxiliary tasks (ablation) | SLADD · FF++ cross-method generalization | AUC (Avg.): 0.709 -> 0.769 | Helped |
Closest papers by meaning
- Transcending Forgery Specificity with Latent Space Augmentation for Generalizable Deepfake Detection (CVPR 2024 · similarity 0.88)
- From Specificity to Generality: Revisiting Generalizable Artifacts in Detecting Face Deepfakes (NeurIPS 2025 · similarity 0.88)
- DF40: Toward Next-Generation Deepfake Detection (NeurIPS 2024 · similarity 0.87)
- Exposing the Deception: Uncovering More Forgery Clues for Deepfake Detection (AAAI 2024 · similarity 0.87)
- DiffusionFake: Enhancing Generalization in Deepfake Detection via Guided Stable Diffusion (NeurIPS 2024 · similarity 0.87)
- Rethinking Vision-Language Model in Face Forensics: Multi-Modal Interpretable Forged Face Detector (CVPR 2025 · similarity 0.87)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.