AI Paper Atlas

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

SWAD: Domain Generalization by Seeking Flat Minima

Junbum Cha, Sanghyuk Chun, Kyungjae Lee, Han-Cheol Cho, Seunghyun Park, Yunsung Lee, Sungrae Park

NeurIPS 2021 · Label-efficient and robust learning › Domain adaptation and generalization · approach: Foundation-model domain generalization · 4 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
apply SWAD on top of ERMERM · 5 DG benchmarks averageout-of-domain accuracy: 63.3 -> 66.9Improved
apply SWAD on top of CORALCORAL · 5 DG benchmarks averageout-of-domain accuracy: 64.5 -> 67.3Improved
apply SWAD on top of SAMSAM · 5 DG benchmarks averageout-of-domain accuracy: 64.5 -> 65.5Improved
apply SWAD on top of CORALCORAL · VLCSout-of-domain accuracy: 78.8 -> 78.9No clear effect
apply SWAD on top of SAMSAM · VLCSout-of-domain accuracy: 79.4 -> 78.5Hurt
remove dense sampling (interval 100 instead of 1) (ablation)SWAD · PACS+VLCS averageout-of-domain accuracy: 82.2 -> 83.0Improved
remove overfit-aware end iteration (average to final iteration) (ablation)SWAD · PACS+VLCS averageout-of-domain accuracy: 82.4 -> 83.0Improved
remove overfit-aware start and end search (fixed 4000-5000) (ablation)SWAD · PACS+VLCS averageout-of-domain accuracy: 81.7 -> 83.0Improved
choose averaging range on validation set instead of overfit-aware (ablation)SWAD · PACS+VLCS averageout-of-domain accuracy: 83.0 -> 82.4Hurt
constant learning rate instead of cyclic for SWA (ablation)SWA · PACS+VLCS averageout-of-domain accuracy: 81.2 -> 81.6No clear effect
SWAD instead of SWA on ImageNetResNet-50 · ImageNet-CmCE: 56.8 -> 55.7Improved
SWAD instead of ERM on background challengeResNet-50 · BGCaccuracy: 8.7 -> 11.8Improved

Later papers that compared with this method

23 later accepted papers report SWAD as a baseline in their results tables; 21 of them report a better number on most of the results they share with it.

15 more not listed.

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