Research map / Label-efficient and robust learning / Recognition and long-tail
Class-Conditional Distribution Balancing for Group Robust Classification
Miaoyun Zhao, Qiang Zhang
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 |
|---|---|---|---|
| use class-conditional distribution balancing instead of class-balancing-only reweighting (ReWeight, BSoftmax, ReWeightCRT) (ablation) | CCDB · Waterbirds worst-group | worst-group accuracy (%): 76.30 -> 90.48 (vs ReWeightCRT) | Improved |
| train biased feature extractor on a smaller data split (gamma 30-50%) instead of full data (ablation) | CCDB stage 1 · cMNIST / Waterbirds | worst-group accuracy | Improved |
| reduce gamma to 10% (ablation) | CCDB stage 1 · Waterbirds | spurious dominance of features | Hurt |
| clip sample-weight logits to [-T, T] (ablation) | CCDB stage 2 · Waterbirds / cCIFAR10 | overall performance | Improved |
| CCDB vs ERM on cMNIST with 0.5% bias-conflicting samples | ERM · cMNIST (0.5%) | test accuracy (%): 35.19 -> 83.20 | Improved |
| CCDB vs ERM on CelebA | ERM · CelebA | worst-group accuracy (%): 47.20 -> 85.27 | Improved |
| CCDB vs ERM on Waterbirds | ERM · Waterbirds | worst-group accuracy (%): 72.60 -> 90.48 | Improved |
| CCDB vs ERM on CivilComments | ERM · CivilComments | worst-group accuracy (%): 66.7 -> 75.00 | Improved |
| CCDB vs bias-supervised GroupDRO on CelebA | GroupDRO · CelebA | worst-group accuracy (%): 88.90 -> 85.27 | Hurt |
| CCDB vs bias-supervised GroupDRO on cCIFAR10 (multi-class) | GroupDRO · cCIFAR10 (0.5%) | test accuracy (%): 33.44 -> 55.07 | Improved |
| apply sample reweighting to features (ablation) | CCDB · feature correlations with bias vs labels | feature spuriousness | Improved |
Closest papers by meaning
- The Group Robustness is in the Details: Revisiting Finetuning under Spurious Correlations (NeurIPS 2024 · similarity 0.89)
- Learning under Localized Minority Imbalance (NeurIPS 2026 · similarity 0.88)
- Balanced Data, Imbalanced Spectra: Unveiling Class Disparities with Spectral Imbalance (ICML 2024 · similarity 0.88)
- Improving Group Robustness on Spurious Correlation Requires Preciser Group Inference (ICML 2024 · similarity 0.87)
- Leveraging Group Classification with Descending Soft Labeling for Deep Imbalanced Regression (AAAI 2025 · similarity 0.87)
- Balancing the Scales: A Theoretical and Algorithmic Framework for Learning from Imbalanced Data (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.
