Research map / Label-efficient and robust learning / Recognition and long-tail
HSFM: Hard-Set-Guided Feature-Space Meta-Learning for Robust Classification under Spurious Correlations
Aryan Yazdan Parast, Khawar Islam, Soyoun Won, Basim Azam, NAVEED AKHTAR
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 |
|---|---|---|---|
| apply HSFM on top of ERM (ResNet-50) | ERM · Waterbirds | worst-group accuracy: 74.6 -> 93.1 | Improved |
| apply HSFM on top of ERM (ResNet-50) | ERM · CelebA | worst-group accuracy: 30.6 -> 89.2 | Improved |
| apply HSFM on top of ERM (ResNet-50) | ERM · Dominoes | worst-group accuracy: 78.6 -> 90.4 | Improved |
| apply HSFM on top of ERM (ResNet-50) | ERM · MetaShift | worst-group accuracy: 64.1 -> 77.2 | Improved |
| apply HSFM on top of ERM (ViT-B/16) | ERM · Waterbirds | worst-group accuracy: 53.9 -> 81.5 | Improved |
| apply HSFM on top of ERM (ViT-B/16) | ERM · MetaShift | worst-group accuracy: 69.7 -> 69.9 | No clear effect |
| apply HSFM on top of ERM (ConvNeXt) | ERM · CelebA | worst-group accuracy: 47.2 -> 81.1 | Improved |
| apply HSFM on top of ERM (ResNet-50) | ERM · CelebA | average accuracy: 95.83 -> 90.6 | Hurt |
| apply HSFM on CLIP linear classifier | ERM · CelebA* | worst-group accuracy: 50.0 -> 81.0 | Improved |
| apply HSFM on pretrained ResNet-50 features | ERM · Stanford Cars | top-1 accuracy: 83.98 -> 85.10 | Improved |
| increase number of inner-loop updates T from 1 or 5 to 10 or more (ablation) | HSFM · Waterbirds | worst-group accuracy | Improved |
| use intermediate support set size (250) rather than larger (ablation) | HSFM · Waterbirds | worst-group accuracy | Improved |
Closest papers by meaning
- On Feature Learning in the Presence of Spurious Correlations (NeurIPS 2022 · similarity 0.87)
- Unraveling Meta-Learning: Understanding Feature Representations for Few-Shot Tasks (ICML 2020 · similarity 0.85)
- Let Samples Speak: Mitigating Spurious Correlation by Exploiting the Clusterness of Samples (CVPR 2025 · similarity 0.85)
- MetaCoCo: A New Few-Shot Classification Benchmark with Spurious Correlation (ICLR 2024 · similarity 0.84)
- Complexity Matters: Feature Learning in the Presence of Spurious Correlations (ICML 2024 · similarity 0.84)
- Decompose-and-Compose: A Compositional Approach to Mitigating Spurious Correlation (CVPR 2024 · similarity 0.83)
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.
