Research map / Label-efficient and robust learning / Recognition and long-tail
Nested Collaborative Learning for Long-Tailed Visual Recognition
Check 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 |
|---|---|---|---|
| Add NIL to the BSCE baseline (ablation) | NCL · CIFAR100-LT IF 100 | accuracy: 50.60 -> 51.24 | Helped |
| Add self-supervision to the BSCE baseline (ablation) | NCL · CIFAR100-LT IF 100 | accuracy: 50.60 -> 51.42 | Helped |
| Add BOD_all to NIL (ablation) | NCL · CIFAR100-LT IF 100 | accuracy: 51.24 -> 52.64 | Helped |
| Add BOD_hard to NIL and BOD_all (ablation) | NCL · CIFAR100-LT IF 100 | accuracy: 52.64 -> 53.19 | Helped |
| Add self-supervision to NIL and NBOD (ablation) | NCL · CIFAR100-LT IF 100 | accuracy: 53.19 -> 53.31 | Helped |
| Evaluate the multiple experts as an ensemble (ablation) | NCL · CIFAR100-LT IF 100 | accuracy: 53.31 -> 54.42 | Helped |
| Remove RandAugment (ablation) | NCL · CIFAR100-LT IF 100 | accuracy: 53.31 -> 47.93 | Hurt |
| Remove balanced probabilities from online distillation (ablation) | NBOD · CIFAR100-LT IF 100 | accuracy: about 1% | Hurt |
| Replace online NBOD with offline distillation (ablation) | NIL+NBOD · CIFAR100-LT IF 100 | accuracy: 1.5% | Hurt |
| Set the hard-category ratio to 0.3 (ablation) | NIL · CIFAR100-LT IF 100 | accuracy | Helped |
| Set the NBOD loss weight to 0.6 (ablation) | NCL · CIFAR100-LT IF 100 | accuracy | Helped |
| Increase the number of experts (ablation) | NCL ensemble · CIFAR100-LT IF 100 | accuracy | Helped |
Closest papers by meaning
- Trust-calibrated Collaborative Learning for Long-Tailed Visual Recognition (CVPR 2026 · similarity 0.87)
- Long-Tailed Multi-Label Visual Recognition by Collaborative Training on Uniform and Re-Balanced Samplings (CVPR 2021 · similarity 0.86)
- Decoupled Optimisation for Long-Tailed Visual Recognition (AAAI 2024 · similarity 0.85)
- Decoupled Contrastive Learning for Long-Tailed Recognition (AAAI 2024 · similarity 0.85)
- SimLTD: Simple Supervised and Semi-Supervised Long-Tailed Object Detection (CVPR 2025 · similarity 0.84)
- Mixture of Nested Experts: Adaptive Processing of Visual Tokens (NeurIPS 2024 · similarity 0.84)
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.