Research map / Image generation / Text-to-image diffusion
Beyond Text Prompts: Precise Concept Erasure through Text-Image Collaboration
Jun Li, Lizhi Xiong, Ziqiang Li, Weiwei Jiang, Zhangjie Fu, Yong Li, Guo-Sen Xie
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 continuous convex concept manifold (ablation) | TICoE · Erase gun | UDA: 0.38 -> 0.02 | Improved |
| add continuous convex concept manifold (ablation) | TICoE · Erase gun | FID: 30.41 -> 30.86 | No clear effect |
| add hierarchical visual representation learning (ablation) | TICoE · Erase gun | UDA: 0.16 -> 0.02 | Improved |
| add hierarchical visual representation learning (ablation) | TICoE · Erase gun | CLIP: 0.285 -> 0.304 | Improved |
| use full manifold instead of prompt bank of 10 (ablation) | TICoE · Erase gun | UDA: 0.26 -> 0.02 | Improved |
| use full manifold instead of prompt bank of 20 (ablation) | TICoE · Erase gun | UDA: 0.12 -> 0.02 | Improved |
| enlarge prompt bank to 40-50 (ablation) | TICoE · Erase gun | UDA: 0.20 -> 0.22 | Hurt |
| use two scales {1.0, 0.75} instead of default (ablation) | TICoE · Erase gun | UDA: 0.26 -> 0.02 | Hurt |
| use four scales instead of default (ablation) | TICoE · Erase gun | FID: 32.74 -> 30.86 | Hurt |
| use text-image collaborative erasure versus text-only ESD | TICoE · Erase gun | UDA: 0.20 -> 0.02 | Improved |
| add image guidance versus Co-Erasing | TICoE · Erase gun, preservation | MCP (camera): 39.68% -> 92.06% | Improved |
| use 200 instance images versus 100 (ablation) | TICoE · Erase gun, hyperparameter | MCP: 74.6% -> 92.1% | Improved |
Closest papers by meaning
- One Image is Worth a Thousand Words: A Usability Preservable Text-Image Collaborative Erasing Framework (ICML 2025 · similarity 0.94)
- Circumventing Concept Erasure Methods For Text-To-Image Generative Models (ICLR 2024 · similarity 0.91)
- Semantic Surgery: Zero-Shot Concept Erasure in Diffusion Models (NeurIPS 2025 · similarity 0.91)
- Localized Concept Erasure for Text-to-Image Diffusion Models Using Training-Free Gated Low-Rank Adaptation (CVPR 2025 · similarity 0.91)
- MACE: Mass Concept Erasure in Diffusion Models (CVPR 2024 · similarity 0.90)
- Precise, Fast, and Low-cost Concept Erasure in Value Space: Orthogonal Complement Matters (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.
