Research map / Label-efficient and robust learning / Recognition and long-tail
A Fine-Grained Analysis on Distribution Shift
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 |
|---|---|---|---|
| Use CycleGAN learned augmentation instead of the ResNet baseline | ResNet · Spurious correlation, across evaluated datasets | top-1 accuracy | Helped |
| Use ImageNet pretraining instead of the ResNet baseline | ResNet · Low-data drift, across evaluated datasets | top-1 accuracy | Helped |
| Use ImageNet pretraining instead of the ResNet baseline | ResNet · Unseen data shift, across evaluated datasets | top-1 accuracy | Helped |
| Use ImageNet pretraining | ResNet · Multiple shifts and datasets | top-1 accuracy | Mixed |
| Use ImageNet augmentation instead of the ResNet baseline | ResNet · Multiple shifts and datasets | top-1 accuracy | Helped |
| Use RandAugment, AugMix, or AutoAugment instead of the ResNet baseline | ResNet · Multiple shifts and datasets | top-1 accuracy | Mixed |
| Use learned augmentation instead of the ResNet baseline | ResNet · Spurious correlation, low-data drift, and unseen data shift | top-1 accuracy | Helped |
| Use domain generalization methods instead of the ResNet baseline | ResNet · Low-data drift and unseen data shift | top-1 accuracy | Mixed |
| Use JTT or BN-Adapt instead of the ResNet baseline | ResNet · Evaluated distribution shifts | top-1 accuracy | No clear effect |
| Use heuristic augmentation when reducing the training dataset size | ResNet · Dataset size condition | top-1 accuracy | Hurt |
| Use pretraining or learned augmentation when reducing the training dataset size | ResNet · Dataset size condition | top-1 accuracy | Helped |
| Add invert augmentation (ablation) | RandAugment · Unseen data shift, dSprites and Shapes3D | top-1 accuracy | Helped |
Closest papers by meaning
- "What is Different Between These Datasets?" A Framework for Explaining Data Distribution Shifts (NeurIPS 2026 · similarity 0.89)
- An Information-theoretic Approach to Distribution Shifts (NeurIPS 2021 · similarity 0.89)
- Estimating and Explaining Model Performance When Both Covariates and Labels Shift (NeurIPS 2022 · similarity 0.89)
- Generalization vs Specialization under Concept Shift (NeurIPS 2025 · similarity 0.87)
- Multiply Robust Estimation for Local Distribution Shifts with Multiple Domains (ICML 2024 · similarity 0.87)
- When to retrain a machine learning model (ICML 2025 · similarity 0.87)
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.