AI Paper Atlas

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

A Broad Study of Pre-training for Domain Generalization and Adaptation

Donghyun Kim, Kaihong Wang, Stan Sclaroff, Kate Saenko

ECCV 2022 · Label-efficient and robust learning › Domain adaptation and generalization · approach: Adversarial domain adaptation · 78 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 Swin-T instead of ResNet-50 (both ImageNet-1K)single-source DG · OH/CUB/WILD/DN avgaccuracy: 45.6 -> 49.3Improved
use Swin-S instead of ResNet-101single-source DG · OH/CUB/WILD/DN avgaccuracy: 48.5 -> 52.8Improved
use DeiT-S instead of ResNet-50single-source DG · CUBaccuracy: 42.3 -> 58.0Improved
use ConvNeXt-T instead of ResNet-50single-source DG · CUBaccuracy: 42.3 -> 42.5No clear effect
pretrain on ImageNet-22K instead of ImageNet-1KConvNeXt-B · CUBaccuracy: +19.8%Improved
pretrain with ALBEF image-text data instead of ImageNet-supervisedViT-B ALBEF · CUBaccuracyHurt
pretrain with ALBEF image-text dataViT-B ALBEF · DomainNetaccuracy: 45.3 -> 50.3Improved
use MoCo self-supervised pretraining instead of supervised (ablation)ResNet-50 · Office-Homeaccuracy: 64.1 -> 53.7Hurt
use DINO self-supervised pretraining instead of supervised (ablation)ResNet-50 · DomainNetaccuracy: 30.2 -> 23.1Hurt
add BEiT self-supervised learning to supervised pretraining (ablation)ViT-L ImageNet-22K · CUBaccuracy: 76.5 -> 78.4Improved
add BEiT self-supervised learning to supervised pretraining (ablation)ViT-L ImageNet-22K · Office-Homeaccuracy: 82.3 -> 80.9Hurt
fine-tune the text encoder in ALBEF (ablation)ALBEF · Office-Homeaccuracy: 80.4 -> 79.3Hurt

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