AI Paper Atlas

Research map / Architectures and efficiency / RNNs and normalization

Certified Circuits: Stability Guarantees for Mechanistic Circuits

Alaa Anani, Tobias Lorenz, Bernt Schiele, Mario Fritz, Jonas Fischer

ICML 2026 · Architectures and efficiency › RNNs and normalization · approach: Mechanistic circuit interpretability · 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
add certification with abstention on unstable componentstop-K relevance circuits · ImageNet ID (ResNet-101)peak cACC / size K: 0.83 -> 0.95 cACC; K 0.700 -> 0.336Improved
add certification with abstentiontop-K relevance circuits · ImageNet-A (ResNet-101)peak cACC: 0.60 -> 0.94Improved
add certification with abstentiontop-K relevance circuits · OOD-CV (ResNet-101)peak cACC: 0.73 -> 0.93Improved
add certification with abstentionEAP-IG circuits · IOI, IOI-Hard, Greater-Than (ID)peak cACC / edges: 58%, 38%, 75% fewer edges at 1.00 cACCImproved
add certification with abstentionEAP-IG circuits · IOI-Hard (OOD)peak cACC: 0.85 -> 0.91Improved
add certification with abstentionEAP-IG circuits · ID language taskspeak cACC: 1.00 -> 1.00No clear effect
use activation scorer instead of relevance for certified circuits (ablation)Certified Circuits · ImageNet-A (ResNet-101)peak cACC: 0.94 -> 0.36Hurt
use rank scorer instead of relevance for certified circuits (ablation)Certified Circuits · ImageNet (ResNet-101)circuit size reduction: down 52% -> down 17%Hurt
increase certified radius to r=42, 50, 59 on language tasks (ablation)Certified EAP-IG · GPT-2 Small IOI, IOI-Hard, Greater-ThancACCNo clear effect
increase deletion probability for larger radiusCertified Circuits · small concept datasetsbase algorithm qualityHurt
add certification to top-K circuits, rediscovering on shifted dataCertified Circuits · ImageNet vs OOD datasetsIoU of circuit structureImproved
certification across random seedsCertified Circuits · ImageNet 100 classespairwise IoU: mean IoU 0.973Improved

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