AI Paper Atlas

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

Domain Generalization by Mutual-Information Regularization with Pre-trained Models

Junbum Cha, Kyungjae Lee, Sungrae Park, Sanghyuk Chun

ECCV 2022 · Label-efficient and robust learning › Domain adaptation and generalization · approach: Optimization-based domain generalization · 153 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 MIRO mutual-information regularization to ERMERM ResNet-50 ImageNet · DomainBed averageout-of-domain accuracy (%): 64.2 -> 65.9Improved
add MIRO to ERMERM RegNetY-16GF SWAG · DomainBed averageout-of-domain accuracy (%): 68.0 -> 74.1Improved
add MIRO to ERMERM ViT-B CLIP · DomainBed averageout-of-domain accuracy (%): 61.1 -> 73.7Improved
add MIRO to ERMERM ResNet-50 CLIP · DomainBed averageout-of-domain accuracy (%): 44.9 -> 61.2Improved
add MIRO to ERMERM Barlow Twins · DomainBed averageout-of-domain accuracy (%): 58.4 -> 61.9Improved
add MIRO to SWADERM + SWAD · DomainBed averageout-of-domain accuracy (%): 66.9 -> 68.1Improved
add MIRO to SWAD with RegNetERM + SWAD RegNetY-16GF · DomainBed averageout-of-domain accuracy (%): 73.2 -> 77.3Improved
add MIRO to ERMERM MoCo v3 · PACSout-of-domain accuracy (%): 86.7 -> 86.3Hurt
naively fine-tune CLIP ViT-B instead of ImageNet ResNet-50ERM · DomainBed averageout-of-domain accuracy (%): 64.2 -> 61.1Hurt
add MIRO to ERMERM ResNet-50 · Camelyon17 averageaccuracy (%): 94.9 -> 95.6Improved
add pre-trained weights (ERM+ vs ERM-) (ablation)ERM ResNet-50 · PACSaccuracy (%): 51.6 -> 84.2Improved
use class-conditional MI (C-MIRO) instead of MIRO (ablation)MIRO · DomainBed 4 datasetsaccuracy (%): 71.3 -> 71.0No clear effect

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