Research map / Image generation / Text-to-image diffusion
CE-SDWV: Effective and Efficient Concept Erasure for Text-to-Image Diffusion Models via a Semantic-Driven Word Vocabulary
Jiahang Tu, Qian Feng, Jiahua Dong, Hanbin Zhao, Chao Zhang, Nicu Sebe, Hui Qian
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 |
|---|---|---|---|
| increase top-k principal components removed from 3 to 5 (ablation) | CE-SDWV · I2P NudeNet | total detections: 243 -> 84 | Improved |
| increase top-k from 1 to 5 (ablation) | CE-SDWV · I2P NudeNet | total detections: 314 -> 84 | Improved |
| increase top-k from 5 to 9 (ablation) | CE-SDWV · MS-COCO 30K | FID: 13.66 -> 14.21 | Hurt |
| increase top-k from 5 to 9 (ablation) | CE-SDWV · MS-COCO 30K | CLIP score: 30.80 -> 30.43 | Hurt |
| increase semantic matrix size N from 50 to 4000 tokens (ablation) | CE-SDWV · MS-COCO 30K | FID: 15.89 -> 13.82 | Improved |
| increase semantic matrix size N from 50 to 4000 tokens (ablation) | CE-SDWV · MS-COCO 30K | CLIP score: 30.25 -> 30.69 | Improved |
| add token optimization (orthogonal) after suppression (ablation) | CE-SDWV · MS-COCO 30K | FID: 16.89 -> 13.66 | Improved |
| add token optimization (orthogonal) after suppression (ablation) | CE-SDWV · I2P NudeNet | total detections: 78 -> 84 | Hurt |
| add orthogonal gradient projection to token optimization (ablation) | CE-SDWV · I2P NudeNet | total detections: 134 -> 84 | Improved |
| add orthogonal gradient projection to token optimization (ablation) | CE-SDWV · MS-COCO 30K | CLIP score: 30.98 -> 30.80 | Hurt |
| compare with ESD on style erasure retention | CE-SDWV vs ESD · UnlearnCanvas style erasure | IRA: 80.97% -> 98.62% | Improved |
| compare with UCE under UnlearnDiffAtk | CE-SDWV vs UCE · UnlearnDiffAtk on I2P | ASR (%): 79.58 -> 8.47 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 2 of 2 on most shared results.
- SalUn — SalUn: Empowering Machine Unlearning via Gradient-based Weight Saliency in Both Image Classification and Generation (ICLR 2024) · better on 5 of 6 shared results
- Receler — Receler: Reliable Concept Erasing of Text-to-Image Diffusion Models via Lightweight Erasers (ECCV 2024) · better on 1 of 1 shared result
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- Semantic Surgery: Zero-Shot Concept Erasure in Diffusion Models (NeurIPS 2025 · similarity 0.92)
- Growth Inhibitors for Suppressing Inappropriate Image Concepts in Diffusion Models (ICLR 2025 · similarity 0.91)
- MACE: Mass Concept Erasure in Diffusion Models (CVPR 2024 · similarity 0.90)
- Comprehensive Assessment and Analysis for NSFW Content Erasure in Text-to-Image Diffusion models (NeurIPS 2025 · similarity 0.90)
- Localized Concept Erasure for Text-to-Image Diffusion Models Using Training-Free Gated Low-Rank Adaptation (CVPR 2025 · similarity 0.90)
- All but One: Surgical Concept Erasing with Model Preservation in Text-to-Image Diffusion Models (AAAI 2024 · similarity 0.90)
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.
