Research map / Label-efficient and robust learning / Recognition and long-tail
Disentangling Sampling and Labeling Bias for Learning in Large-output Spaces
Ankit Singh Rawat, Aditya K Menon, Wittawat Jitkrittum, Sadeep Jayasumana, Felix Yu, Sashank Reddi, Sanjiv Kumar
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 within-batch negatives with constant weighting | softmax cross-entropy · CIFAR-100-LT | balanced error on Tail classes | Improved |
| Use within-batch negatives with constant weighting | softmax cross-entropy · long-tail benchmarks | balanced error on Tail classes | Improved |
| Use within-batch negatives with relative weighting instead of constant weighting (ablation) | softmax cross-entropy · long-tail benchmarks | balanced error on Head classes | Improved |
| Use the proposed tail weighting | uniform and within-batch negative sampling · long-tail benchmarks | balanced error on Tail classes | Improved |
| Replace the full-label logit-adjustment loss with tail-weighted negative sampling | softmax cross-entropy · long-tail benchmarks | balanced error on Tail classes | Hurt |
| Use within-batch negatives | softmax cross-entropy · ImageNet-LT | balanced error on Tail classes | Improved |
| Use uniform negatives instead of within-batch negatives (ablation) | softmax cross-entropy · retrieval benchmarks | Recall@k on Tail | Hurt |
| Use within-batch negatives instead of uniform negatives (ablation) | softmax cross-entropy · retrieval benchmarks | Recall@k on Head | Hurt |
| Use uniform negatives instead of within-batch negatives (ablation) | softmax cross-entropy · retrieval benchmarks | Recall@k on Head | Improved |
| Use negative sampling with contrastive loss instead of the standard baseline (ablation) | contrastive loss · long-tail benchmarks | balanced error on Tail classes | Hurt |
| Reduce the number of sampled negatives (ablation) | softmax cross-entropy · ImageNet-LT | balanced error | Hurt |
Closest papers by meaning
- Rethinking the Value of Labels for Improving Class-Imbalanced Learning (NeurIPS 2020 · similarity 0.90)
- Distribution-Balanced Loss for Multi-Label Classification in Long-Tailed Datasets (ECCV 2020 · similarity 0.88)
- Towards Unbiased Exploration in Partial Label Learning (JMLR 2024 · similarity 0.88)
- Long-Tailed Partial Label Learning via Dynamic Rebalancing (ICLR 2023 · similarity 0.88)
- Semi-supervised Multi-label Learning with Balanced Binary Angular Margin Loss (NeurIPS 2024 · similarity 0.87)
- Rebalancing Multi-Label Class-Incremental Learning (AAAI 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.
