AI Paper Atlas

Research map / Image generation / Text-to-image diffusion

OrthoEraser: Coupled-Neuron Orthogonal Projection for Concept Erasure

Chuancheng Shi, Wenhua Wu, Fei Shen, Xiaogang Zhu, Kun Hu, Zhiyong Wang

ECCV 2026 · Image generation › Text-to-image diffusion · approach: Concept erasure in diffusion models · 8 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
select the global optimal layer 10 instead of layer 9 (ablation)OrthoEraser · I2P nuditytotal detections: 17 -> 5Improved
select the global optimal layer instead of first sensitivity peak layer 3 (ablation)OrthoEraser · I2P nuditytotal detections: 24 -> 5Improved
suppress identified sensitive neurons instead of random neurons of same size (ablation)OrthoEraser · I2P nuditytotal detections: 598 -> 5Improved
add coupled-neuron orthogonal projection to sensitive suppression (ablation)OrthoEraser · I2P nuditytotal detections: 17 -> 5Improved
add coupled-neuron orthogonal projection to sensitive suppression (ablation)OrthoEraser · MS COCO-30KFID: 16.64 -> 1.15Improved
add coupled-neuron orthogonal projection to sensitive suppression (ablation)OrthoEraser · MS COCO-30KCLIP score: 30.87 -> 31.33Improved
add sensitive-neuron suppression to coupled-only suppression (ablation)OrthoEraser · I2P nuditytotal detections: 604 -> 5Improved
use orthogonal projection instead of coupled-aligned suppression (ablation)OrthoEraser · MS COCO-30KFID: 23.95 -> 1.15Improved
use orthogonal projection instead of coupled-aligned suppression (ablation)OrthoEraser · MS COCO-30KCLIP score: 26.31 -> 31.33Improved
apply OrthoEraser vs SNCE on nuditySNCE baseline · I2P nuditytotal detections: 17 -> 5Improved
extend to violence conceptOrthoEraser · I2P Violencedetection rate: 40.1 -> 15.6Improved
apply to adversarial prompts Ring-A-BellOrthoEraser · Ring-A-Bellattack success rate: 98.7 -> 2.7Improved

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 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