AI Paper Atlas

Research map / Image generation / Fast diffusion sampling

Data Circuit Breaker: Identifying Training, Test, and Generated Data in Image Generative Models

Bihe Zhao, Michel Meintz, Juangui Xu, Franziska Boenisch, Adam Dziedzic

ECCV 2026 · Image generation › Fast diffusion sampling · approach: Distributional control of diffusion

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
add autoencoder filtering and cross-signal stages (DCB) over ICASIAR MIA · MGI direct training IARsTPR@1%FPR: 52.6 -> 87.4Improved
add DCB over CLiD for diffusion modelsDM MIA · MGI direct training DMsTPR@1%FPR: 39.7 -> 77.8Improved
add DCB cross-generator attribution over ICASVAR · MGI model derivativeTPR@1%FPR: 81.3 -> 95.8Improved
add DCB over ICASStable Diffusion 2.1 · MGI model derivativeTPR@1%FPR: 64.4 -> 83.7Improved
add DCB over Delta for memorized samplesRAR-XXL · Memorized training vs regeneratedAUC: 61.8 -> 97.5Improved
add DCB over RMIA with shadow modelsRAR-XXL · MGI model derivativeTPR@1%FPR: 82.7 -> 96.7Improved
add DCB over RMIA with shadow modelsVAR-d30 · MGI model derivativeTPR@1%FPR: 76.4 -> 95.8Improved
add augmentation to encoder fine-tuning (ablation)DCB Stage 1 · Adversarial attack eps=1TPR@1%FPR: 68.7 -> 97.4Improved
add augmentation to encoder fine-tuning (ablation)DCB Stage 1 · JPEG (60) and Resize (0.5)TPR@1%FPR: JPEG 91.7 -> 96.1; Resize 88.5 -> 98.4Improved
replace ground-truth prompts with BLIP2 or LLaVA captions (ablation)DCB · SD1.4 derivativeTPR@1%FPR: 82.6 -> 81.5; 82.6 -> 78.7No clear effect
replace ground-truth prompts with BLIP2 or LLaVA captions (ablation)ICAS · SD1.4 derivativeTPR@1%FPR: 63.5 -> 46.2; 63.5 -> 33.6Hurt
DCB versus ICAS on natural members vs non-membersRAR · MGI direct trainingTPR@1%FPR: 72.5 -> 72.5No 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