AI Paper Atlas

Research map / Label-efficient and robust learning / Domain adaptation and generalization

Self-Challenging Improves Cross-Domain Generalization

Zeyi Huang, Haohan Wang, Eric P. Xing, Dong Huang

ECCV 2020 · Label-efficient and robust learning › Domain adaptation and generalization · approach: Augmentation and mixture-of-experts generalization

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
select dropped features by gradient instead of random (ablation)Spatial RSC · PACS (ResNet18)average accuracy (%): 81.05 -> 82.91Improved
select dropped features by gradient instead of activation (ablation)Spatial RSC · PACS (ResNet18)average accuracy (%): 82.03 -> 82.91Improved
lower dropping percentage from 50% to 33.3% (ablation)Spatial RSC · PACS (ResNet18)average accuracy (%): 82.91 -> 83.95Improved
raise dropping percentage to 66.7% (ablation)Spatial RSC · PACS (ResNet18)average accuracy (%): 83.95 -> 81.97Hurt
reduce batch percentage from 50% to 33.3% (ablation)Spatial RSC · PACS (ResNet18)average accuracy (%): 83.95 -> 84.17No clear effect
add channel-wise RSC to spatial-wise RSC (ablation)RSC · PACS (ResNet18)average accuracy (%): 84.17 -> 85.15Improved
train with RSC instead of baselineResNet18 · PACSaverage accuracy (%): 79.94 -> 85.15Improved
train with RSC instead of baselineResNet50 · PACSaverage accuracy (%): 83.29 -> 87.83Improved
train with RSC instead of baselineAlexNet · VLCSaverage accuracy (%): 72.76 -> 75.43Improved
train with RSC instead of baselineResNet18 · Office-Homeaverage accuracy (%): 60.51 -> 63.12Improved
train with RSC instead of baselineAlexNet · ImageNet-Sketchaverage accuracy (%): 12.04 -> 16.12Improved
train with RSC instead of baselineResNet50 · ImageNet valaverage accuracy (%): 76.13 -> 77.18Improved

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.

Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.

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