AI Paper Atlas

Research map / Image generation / Autoregressive and latent generation

GUDA: Counterfactual Group-wise Training Data Attribution for Diffusion Models via Unlearning

Naoki Murata, Yuhta Takida, Chieh-Hsin Lai, Toshimitsu Uesaka, Bac Nguyen, Stefano Ermon, Yuki Mitsufuji

ICML 2026 · Image generation › Autoregressive and latent generation · approach: Synthetic training data from generators · 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
use ReTrack forget loss instead of ESD in the unlearning operator (ablation)GUDA · CIFAR-10 group attributionTop-1 agreement: 0.619 -> 0.727Improved
replace CLIP-similarity attribution with unlearning-based counterfactual attributionGUDA vs CLIPA · CIFAR-10 group attributionTop-1 agreement: 0.662 -> 0.727Improved
use unlearning approximation instead of LOGO retrainingGUDA · CIFAR-10 group attributiontotal wall-clock time (h:mm): 207:47 -> 2:02Improved
aggregate instance-level gradient scores (TRAK) to group levelTRAK · CIFAR-10 group attributionTop-1 agreement: 0.118 (10% chance)Hurt
GUDA instead of DAS in head metrics, with faster query timeGUDA · CIFAR-10 group attributionTop-1 agreement / total time: 0.716 -> 0.727; 35:24 -> 2:02No clear effect
use GUDA-C anchor redirection instead of CLIPA similarityGUDA-C · UnlearnCanvas (16 styles)Top-1 agreement: 0.338 -> 0.456Improved
instance-level unlearning of a synthesized image (Wang et al.) for group attributionWang et al. (2024) · UnlearnCanvas (16 styles)Top-1 agreement: 0.047Hurt
use CLIP-weighted instead of uniform retain-style anchor selection (ablation)GUDA-C · UnlearnCanvas (16 styles)Spearman: 0.104 -> 0.239Improved
use retain-style anchor instead of style-removed condition (ablation)GUDA-C · UnlearnCanvas (16 styles)Top-1 agreement: 0.400 -> 0.456Improved
use GUDA-C instead of CLIPA on NDCG@3GUDA-C · UnlearnCanvas (16 styles)NDCG@3: 0.672 -> 0.734Improved
use GUDA-C unlearning instead of LOGO retrainingGUDA-C · UnlearnCanvas (16 styles)total wall-clock time (h:mm): 46:08 -> 8:54Improved

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