Research map / Image generation / Autoregressive and latent generation
FAIR: Feature-Augmented Implicit Regularization for AI-generated Fake Image Detection
Md Redwanul Haque, Manzur Murshed, Manoranjan Paul, Tsz-Kwan Lee
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (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 |
|---|---|---|---|
| add FAIR (SCS prior during training) | PatchCraft · GenImage mean | accuracy (%): 82.36 -> 90.40 | Improved |
| add FAIR | AIDE · GenImage mean | accuracy (%): 86.88 -> 91.01 | Improved |
| add FAIR | AIDE · AIGCDetect mean | accuracy (%): 93.02 -> 92.14 | Hurt |
| add FAIR | PatchCraft · Fake2M mean | accuracy (%): 80.46 -> 78.82 | Hurt |
| add FAIR | AIDE · Fake2M mean | accuracy (%): 69.12 -> 76.18 | Improved |
| add FAIR | PatchCraft · DRCT-2M mean | accuracy (%): 71.39 -> 74.39 | Improved |
| add FAIR | AIDE · DRCT-2M mean | accuracy (%): 66.68 -> 68.83 | Improved |
| add FAIR | AIDE · GenImage BigGAN | accuracy (%): 66.89 -> 78.18 | Improved |
| add FAIR | AIDE · GenImage source domain SDv1.4 | accuracy (%): 99.74 -> 99.63 | No clear effect |
| add dropout p=0.5 (ablation) | AIDE · GenImage mean | accuracy (%): 86.9 -> 86.5 | No clear effect |
| add L1 penalty 1e-2 (ablation) | AIDE · GenImage mean | accuracy (%): 86.9 -> 50.0 | Hurt |
| add L2 penalty 1e-2 (ablation) | AIDE · GenImage mean | accuracy (%): 86.9 -> 85.0 | Hurt |
Closest papers by meaning
- A Bias-Free Training Paradigm for More General AI-generated Image Detection (CVPR 2025 · similarity 0.86)
- Community Forensics: Using Thousands of Generators to Train Fake Image Detectors (CVPR 2025 · similarity 0.86)
- Orthogonal Subspace Decomposition for Generalizable AI-Generated Image Detection (ICML 2025 · similarity 0.86)
- Few-Shot Learner Generalizes Across AI-Generated Image Detection (ICML 2025 · similarity 0.86)
- Breaking Semantic Artifacts for Generalized AI-generated Image Detection (NeurIPS 2024 · similarity 0.85)
- BIAS-ID: A Framework for Analyzing Transformation Biases in AI-Generated Image Detectors (NeurIPS 2026 · similarity 0.85)
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.
