AI Paper Atlas

Research map / Adversarial robustness and security / Certified and empirical robustness

Robustness via Cross-Domain Ensembles

Teresa Yeo, Oğuzhan Fatih Kar, Amir Zamir

ICCV 2021 · Adversarial robustness and security › Certified and empirical robustness · approach: Robustness-accuracy trade-off analysis · 32 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 8 diverse middle-domain paths with simple probability averaging instead of a single ResNet-50ResNet-50 classifier · ImageNet-CmCE: 76.21 -> 67.85Improved
Replace same-architecture deep ensemble members (different inits) with middle-domain paths8-path ResNet-50 ensemble · ImageNet-CmCE: 70.43 -> 67.85Improved
Replace deep ensemble with middle-domain ensemble8-path ResNet-50 ensemble · ImageNet (clean)clean error: 21.50 -> 21.61No clear effect
Use an ensemble of 8 paths instead of a single ResNet-50ResNet-50 classifier · ImageNet (clean)clean error: 24.37 -> 21.61Improved
Add inverse variance merging instead of uniform merging (ablation)Cross-domain ensemble · Taskonomy with Common Corruptions (normals, depth, reshading)l1 errorImproved
Add inverse variance merging instead of network (stacking) merging (ablation)Cross-domain ensemble · Taskonomy with Common Corruptionsl1 errorNo clear effect
Add sigma training (ST) and cross-task consistency (X-TC) (ablation)Cross-domain ensemble · Taskonomy with Common Corruptionsl1 error vs deep ensemblesNo clear effect
Increase number of paths (middle domains) in the ensemble (ablation)Cross-domain ensemble vs deep ensembles · Taskonomy with Common Corruptionsaverage l1 errorImproved
Use middle domains instead of random inits in deep ensembles that also have ST, X-TC and merging (ablation)Deep ensembles with ST and X-TC · Taskonomy with Common Corruptionsl1 errorImproved
Use the proposed middle-domain ensemble instead of a deep ensemble under I-FGSM attacksCross-domain ensemble · Taskonomy with adversarial I-FGSM attacksprediction errorImproved
Use a different choice of middle domain sets (ablation)Cross-domain ensemble · Taskonomy with Common Corruptionsaverage l1 errorNo clear effect
Add sigma training so uncertainty tracks error under shift (ablation)Per-path sigma estimates · Taskonomy epistemic uncertainty vs errorsigma vs error correlationImproved

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