Research map / Image generation / Text-to-image diffusion
Reliable and Efficient Concept Erasure of Text-to-Image Diffusion Models
Chao Gong, Kai Chen, Zhipeng Wei, Jingjing Chen, Yu-Gang Jiang
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 |
|---|---|---|---|
| derive and erase new embeddings iteratively instead of single-pass UCE-style edit | UCE · I2P nudity | total nude parts detected: 186 -> 66 | Improved |
| apply RECE instead of UCE against Ring-A-Bell | UCE · Ring-A-Bell | attack success rate (%): 33.10 -> 13.38 | Improved |
| RECE versus CA under white-box P4D attack | RECE · P4D | attack success rate (%): 60.56 -> 64.79 | Hurt |
| RECE vs UCE on COCO-30k image quality | UCE · COCO-30k | FID: 1.82 -> 2.82 | Hurt |
| run derive-erase for more epochs (epochs 3-4) | RECE · nudity erasure | model generation ability | Hurt |
| increase regularization coefficient lambda | RECE · COCO-30k validation subset | CLIP score | Improved |
| increase regularization coefficient lambda | RECE · I2P | nude parts detected | No clear effect |
| apply RECE for male nudity erasure | RECE · I2P | female/male nudity ratio: female 204 -> 8, male 56 -> 10 | Mixed |
| apply RECE for artist style removal compared to UCE | UCE · Erase Van Gogh | LPIPS_d: 0.20 -> 0.23 | Improved |
| apply RECE for object removal compared to UCE | UCE · Imagenette object removal | accuracy of erased class (%): 2.6 -> 0.3 | Improved |
Closest papers by meaning
- Localized Concept Erasure for Text-to-Image Diffusion Models Using Training-Free Gated Low-Rank Adaptation (CVPR 2025 · similarity 0.93)
- All but One: Surgical Concept Erasing with Model Preservation in Text-to-Image Diffusion Models (AAAI 2024 · similarity 0.93)
- MACE: Mass Concept Erasure in Diffusion Models (CVPR 2024 · similarity 0.93)
- Precise, Fast, and Low-cost Concept Erasure in Value Space: Orthogonal Complement Matters (CVPR 2025 · similarity 0.92)
- Semantic Surgery: Zero-Shot Concept Erasure in Diffusion Models (NeurIPS 2025 · similarity 0.92)
- Concept Pinpoint Eraser for Text-to-image Diffusion Models via Residual Attention Gate (ICLR 2025 · similarity 0.92)
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.
