Research map / Image generation / Text-to-image diffusion
Neighbor-Aware Localized Concept Erasure in Text-to-Image Diffusion Models
Zhuan Shi, Alireza Dehghanpour Farashah, Rik de Vries, Golnoosh Farnadi
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 Stage 2 and Stage 3 (attention gating, hard clean-up) to Stage 1 (ablation) | NLCE · Oxford Flowers | Acct, Ho | Improved |
| Add Stage 2 and Stage 3 to Stage 1 (ablation) | NLCE · Oxford Flowers | CLIP score | Hurt |
| Add Stage 3 after Stage 2 (ablation) | NLCE · Stanford Dogs | erasure metrics | No clear effect |
| Add Stage 2 and 3 to Stage 1 (ablation) | NLCE · Celebrity | LPIPSu / identity leakage | Mixed |
| Neighbor-aware erasure (NLCE) instead of GLoCE | localized erasure · Stanford Dogs: Bluetick | Ho: 72.79 -> 86.31 | Improved |
| Neighbor-aware erasure (NLCE) instead of GLoCE | localized erasure · Oxford Flowers: Camellia | Acct: 92.00 -> 12.00 | Improved |
| Neighbor-aware erasure (NLCE) instead of GLoCE | localized erasure · Celebrity (Elon Musk) | Ho: 97.29 -> 96.55 | No clear effect |
| NLCE instead of GLoCE | concept erasure · 10-dog multi-concept | Acct: 78.00 -> 26.80 | Improved |
| Use NLCE with balanced hyperparameters | NLCE · I2P | nudity removal | Mixed |
| Erase many concepts with training-based methods (MACE, UCE, RECE) | baselines · 10 dogs / 10 flowers | Accr: MACE Accr 1.52 (dogs) | Hurt |
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)
- MACE: Mass Concept Erasure in Diffusion Models (CVPR 2024 · similarity 0.92)
- Semantic Surgery: Zero-Shot Concept Erasure in Diffusion Models (NeurIPS 2025 · similarity 0.91)
- All but One: Surgical Concept Erasing with Model Preservation in Text-to-Image Diffusion Models (AAAI 2024 · similarity 0.90)
- Concept Pinpoint Eraser for Text-to-image Diffusion Models via Residual Attention Gate (ICLR 2025 · similarity 0.90)
- Fine-Grained Erasure in Text-to-Image Diffusion-based Foundation Models (CVPR 2025 · 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.
