AI Paper Atlas

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

IJCV 2026 · Image generation › Text-to-image diffusion · approach: Compositional text-to-image guidance · 10 citations (OpenAlex)

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 changedWhereMeasuredEffect
increase top-k principal components removed from 3 to 5 (ablation)CE-SDWV · I2P NudeNettotal detections: 243 -> 84Improved
increase top-k from 1 to 5 (ablation)CE-SDWV · I2P NudeNettotal detections: 314 -> 84Improved
increase top-k from 5 to 9 (ablation)CE-SDWV · MS-COCO 30KFID: 13.66 -> 14.21Hurt
increase top-k from 5 to 9 (ablation)CE-SDWV · MS-COCO 30KCLIP score: 30.80 -> 30.43Hurt
increase semantic matrix size N from 50 to 4000 tokens (ablation)CE-SDWV · MS-COCO 30KFID: 15.89 -> 13.82Improved
increase semantic matrix size N from 50 to 4000 tokens (ablation)CE-SDWV · MS-COCO 30KCLIP score: 30.25 -> 30.69Improved
add token optimization (orthogonal) after suppression (ablation)CE-SDWV · MS-COCO 30KFID: 16.89 -> 13.66Improved
add token optimization (orthogonal) after suppression (ablation)CE-SDWV · I2P NudeNettotal detections: 78 -> 84Hurt
add orthogonal gradient projection to token optimization (ablation)CE-SDWV · I2P NudeNettotal detections: 134 -> 84Improved
add orthogonal gradient projection to token optimization (ablation)CE-SDWV · MS-COCO 30KCLIP score: 30.98 -> 30.80Hurt
compare with ESD on style erasure retentionCE-SDWV vs ESD · UnlearnCanvas style erasureIRA: 80.97% -> 98.62%Improved
compare with UCE under UnlearnDiffAtkCE-SDWV vs UCE · UnlearnDiffAtk on I2PASR (%): 79.58 -> 8.47Improved

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.

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

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.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy