AI Paper Atlas

Research map / Probabilistic and scientific modeling / Causal discovery

Learning Counterfactually Decoupled Attention for Open-World Model Attribution

Yu Zheng, Boyang Gong, Fanye Kong, Yueqi Duan, Bingyao Yu, Wenzhao Zheng, Lei Chen, Jiwen Lu, Jie Zhou

ICCV 2025 · Probabilistic and scientific modeling › Causal discovery · approach: Causal representation identifiability · 3 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
add factual attention via Causal Expert convolution to CPL (ablation)CPL · OW-DFA Protocol 1 novelARI: 65.71 -> 70.34Improved
add learned counterfactual attention (ablation)CPL + CDAL · OW-DFA Protocol 1 novelARI: 70.34 -> 73.54Improved
add causal attention augmentation (enhanced attention) (ablation)CPL + CDAL · OW-DFA Protocol 1 novelARI: 73.54 -> 76.98Improved
add causal loss (ablation)CPL + CDAL · OW-DFA Protocol 1 novelARI: 68.94 -> 71.75Improved
add decorrelation loss (ablation)CPL + CDAL · OW-DFA Protocol 1 novelARI: 71.75 -> 75.08Improved
add attention diversification loss L_aug (ablation)CPL + CDAL · OW-DFA Protocol 1 novelARI: 75.08 -> 76.98Improved
use 4 causal experts instead of 2, 3, 5 or 6 (ablation)CDAL · OW-DFA Protocol 1 novelARI: 72.57 (N=2) -> 76.98 (N=4) -> 72.00 (N=6)Mixed
replace random counterfactual attention with learned CE-Conv counterfactual attention (ablation)CPL + CDAL · OW-DFA Protocol 1 novelARI: 73.54 -> 76.98Improved
add vanilla attention to CPL without CDAL (ablation)CPL · OW-DFA Protocol 1 novelARI: 65.71 -> 58.81Hurt
plug CDAL into CPLCPL · OW-DFA Protocol 1 novelARI: 65.71 -> 76.98Improved
plug CDAL into POSE for GAN discoveryPOSE · OSMA Unseen Allpurity: 41.04 -> 48.93Improved
plug CDAL into ORCAORCA · OW-DFA Protocol 2 novelARI: 49.15 -> 50.42Mixed

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