AI Paper Atlas

Research map / Image generation / Editing and style transfer

Orthogonal Concept Erasure for Diffusion Models

Yuhao Sun, Lingyun Yu, Hao-Xiang Xu, Fengyuan Miao, Zhuoer Xu, Hongtao Xie

ICML 2026 · Image generation › Editing and style transfer · approach: Inversion-based diffusion editing · 0 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
replace additive parameter update with layer-wise orthogonal update (OCE)editing-based concept erasure · CIFAR-10 object erasure (avg of 5 classes)Acc_e: 8.32 (MACE) -> 4.61Improved
use subspace-level erasure objective with vector-wise preservation instead of vector-wise erasure and preservation (ablation)OCE · Erase 100 celebritiesAcc_e: 7.59 -> 3.44Improved
use subspace-level erasure with vector-wise preservation rather than subspace-level preservation (ablation)OCE · Erase 100 celebritiesHo: 94.22 -> 95.48Improved
subspace-level preservation instead of vector-wise preservation (ablation)OCE · Erase 100 celebritiesFID: 18.10 vs 18.33 FID; Acc_e 4.54 vs 3.44Mixed
add global preservation prior K0 from COCO-30K tokens (ablation)OCE · Erase 100 celebritiesHo: 93.80 -> 95.48Improved
add global preservation prior K0 (ablation)OCE · Erase 100 celebrities (MSCOCO-30k)FID: 22.76 -> 18.33Improved
increase generic concept set for K0 from 1/3 to full (ablation)OCE · Erase 100 celebritiesAcc_e: 4.47 -> 3.44Improved
add adversarial editing (AT) to OCE (ablation)OCE · I2P implicit concept erasureASR: 0.11 -> 0.05Improved
add adversarial editing (AT) to OCE (ablation)OCE · MSCOCO-1K preservationCLIP score: 26.33 -> 26.10Hurt
use OCE instead of MACE for 100-celebrity erasurecelebrity erasure · Erase 100 celebritiesTime (s): 1800 -> 4.3Improved
rotate neurons with a shared layer-wise orthogonal matrix vs neuron-wise rotation (toy experiment) (ablation)SD cross-attention key/value matrices · cat concept generation (qualitative)image qualityHurt
scale weights by alpha=0.5 (magnitude only) (ablation)SD cross-attention key/value matrices · cat concept generation (qualitative)concept semanticsNo clear effect

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