TasksSotADatasetsPapersMethodsSubmitAbout
Papers With Code 2

A community resource for machine learning research: papers, code, benchmarks, and state-of-the-art results.

Explore

Notable BenchmarksAll SotADatasetsPapersMethods

Community

Submit ResultsAbout

Data sourced from the PWC Archive (CC-BY-SA 4.0). Built by the community, for the community.

Papers/Domain Generalization by Mutual-Information Regularization...

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

Junbum Cha, Kyungjae Lee, Sungrae Park, Sanghyuk Chun

2022-03-21Domain Generalization
PaperPDFCode(official)

Abstract

Domain generalization (DG) aims to learn a generalized model to an unseen target domain using only limited source domains. Previous attempts to DG fail to learn domain-invariant representations only from the source domains due to the significant domain shifts between training and test domains. Instead, we re-formulate the DG objective using mutual information with the oracle model, a model generalized to any possible domain. We derive a tractable variational lower bound via approximating the oracle model by a pre-trained model, called Mutual Information Regularization with Oracle (MIRO). Our extensive experiments show that MIRO significantly improves the out-of-distribution performance. Furthermore, our scaling experiments show that the larger the scale of the pre-trained model, the greater the performance improvement of MIRO. Source code is available at https://github.com/kakaobrain/miro.

Results

TaskDatasetMetricValueModel
Domain AdaptationPACSAverage Accuracy96.8MIRO (RegNetY-16GF, SWAD)
Domain AdaptationPACSAverage Accuracy88.4MIRO (ResNet-50, SWAD)
Domain AdaptationOffice-HomeAverage Accuracy83.3MIRO (RegNetY-16GF, SWAD)
Domain AdaptationOffice-HomeAverage Accuracy72.4MIRO (ResNet-50, SWAD)
Domain AdaptationDomainNetAverage Accuracy60.7MIRO (RegNetY-16GF, SWAD)
Domain AdaptationDomainNetAverage Accuracy47MIRO (ResNet-50, SWAD)
Domain AdaptationVLCSAverage Accuracy81.7MIRO (RegNetY-16GF, SWAD)
Domain AdaptationVLCSAverage Accuracy79.6MIRO (ResNet-50, SWAD)
Domain AdaptationTerraIncognitaAverage Accuracy64.3MIRO (RegNetY-16GF, SWAD)
Domain AdaptationTerraIncognitaAverage Accuracy52.9MIRO (ResNet-50, SWAD)
Domain GeneralizationPACSAverage Accuracy96.8MIRO (RegNetY-16GF, SWAD)
Domain GeneralizationPACSAverage Accuracy88.4MIRO (ResNet-50, SWAD)
Domain GeneralizationOffice-HomeAverage Accuracy83.3MIRO (RegNetY-16GF, SWAD)
Domain GeneralizationOffice-HomeAverage Accuracy72.4MIRO (ResNet-50, SWAD)
Domain GeneralizationDomainNetAverage Accuracy60.7MIRO (RegNetY-16GF, SWAD)
Domain GeneralizationDomainNetAverage Accuracy47MIRO (ResNet-50, SWAD)
Domain GeneralizationVLCSAverage Accuracy81.7MIRO (RegNetY-16GF, SWAD)
Domain GeneralizationVLCSAverage Accuracy79.6MIRO (ResNet-50, SWAD)
Domain GeneralizationTerraIncognitaAverage Accuracy64.3MIRO (RegNetY-16GF, SWAD)
Domain GeneralizationTerraIncognitaAverage Accuracy52.9MIRO (ResNet-50, SWAD)

Related Papers

Simulate, Refocus and Ensemble: An Attention-Refocusing Scheme for Domain Generalization2025-07-17GLAD: Generalizable Tuning for Vision-Language Models2025-07-17MoTM: Towards a Foundation Model for Time Series Imputation based on Continuous Modeling2025-07-17InstructFLIP: Exploring Unified Vision-Language Model for Face Anti-spoofing2025-07-16From Physics to Foundation Models: A Review of AI-Driven Quantitative Remote Sensing Inversion2025-07-11Feed-Forward SceneDINO for Unsupervised Semantic Scene Completion2025-07-08Prompt-Free Conditional Diffusion for Multi-object Image Augmentation2025-07-08Integrated Structural Prompt Learning for Vision-Language Models2025-07-08