Research map / Label-efficient and robust learning / Domain adaptation and generalization
Learning to Balance Specificity and Invariance for In and Out of Domain Generalization
Prithvijit Chattopadhyay, Yogesh Balaji, Judy Hoffman
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 changed | Where | Measured | Effect |
|---|---|---|---|
| add domain-specific masks with soft-IoU loss (DMG) | Aggregate baseline, AlexNet · DomainNet out-of-domain | overall accuracy: 28.60 -> 31.15 | Improved |
| add domain-specific masks (DMG) | Aggregate, ResNet-50 · DomainNet out-of-domain | overall accuracy: 44.00 -> 43.63 | No clear effect |
| add domain-specific masks (DMG) | Aggregate*, ResNet-50 · PACS | overall accuracy: 78.38 -> 83.37 | Improved |
| add domain-specific masks (DMG) | Aggregate*, AlexNet · PACS | overall accuracy: 68.38 -> 73.32 | Improved |
| use known-domain mask at test time (DMG-KD) instead of averaging masks | DMG, AlexNet · DomainNet in-domain | overall accuracy: 52.30 -> 54.54 | Improved |
| use known-domain mask at test time (DMG-KD) | DMG, ResNet-18 · DomainNet in-domain | overall accuracy: 59.37 -> 60.59 | Improved |
| use known-domain mask at test time (DMG-KD) | DMG, ResNet-50 · DomainNet in-domain | overall accuracy: 63.79 -> 65.14 | Improved |
| use matching source-domain mask rather than mismatched mask (ablation) | DMG, AlexNet · DomainNet I,P,Q,R,S->C, source Infograph | accuracy: 18.42 -> 23.84 | Improved |
| use matching source-domain mask rather than mismatched mask (ablation) | DMG, AlexNet · DomainNet C,I,P,R,S->Q, source Clipart | accuracy: 59.62 -> 66.70 | Improved |
| average predictions of all masks (Pred-Ens) vs averaging masks (Mask-Ens) (ablation) | DMG, AlexNet · DomainNet out-of-domain | overall accuracy: 31.15 -> 31.10 | No clear effect |
| average predictions of all masks (Pred-Ens) vs averaging masks (Mask-Ens) (ablation) | DMG, AlexNet · DomainNet in-domain | overall accuracy: 52.30 -> 52.17 | No clear effect |
| increase soft-IoU weight lambda_O up to 0.1 and 1 (ablation) | DMG, AlexNet · DomainNet | in-domain and out-of-domain accuracy: only minor variations, slight drop in in-domain at 0.1 and 1 | No clear effect |
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.
- MetaReg — MetaReg: Towards Domain Generalization using Meta-Regularization (NeurIPS 2018) · better on 1 of 1 shared result
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
- Exploiting Domain-Specific Features to Enhance Domain Generalization (NeurIPS 2021 · similarity 0.91)
- Disentangling Masked Autoencoders for Unsupervised Domain Generalization (ECCV 2024 · similarity 0.91)
- Failure Modes of Domain Generalization Algorithms (CVPR 2022 · similarity 0.91)
- Deep Domain Generalization via Conditional Invariant Adversarial Networks (ECCV 2018 · similarity 0.90)
- Learning Informative Invariant Representations via Hierarchical Latent Decomposition (NeurIPS 2026 · similarity 0.90)
- Robust Domain Generalization under Divergent Marginal and Conditional Distributions (NeurIPS 2026 · similarity 0.89)
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.
