Research map / Label-efficient and robust learning / Recognition and long-tail
Invariant Feature Learning for Generalized Long-Tailed Classification
Kaihua Tang, Mingyuan Tao, Jiaxin Qi, Zhenguang Liu, Hanwang 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 |
|---|---|---|---|
| add IFL (two environments with invariant metric loss) (ablation) | ResNeXt-50 baseline · ImageNet-GLT GLT protocol | accuracy: 34.75 -> 37.96 | Improved |
| add IFL | ResNeXt-50 baseline · ImageNet-GLT CLT protocol | accuracy: 42.52 -> 45.97 | Improved |
| add IFL | ResNeXt-50 baseline · ImageNet-GLT ALT protocol | accuracy: 41.73 -> 45.89 | Improved |
| add IFL | RandAug · ImageNet-GLT GLT protocol | accuracy: 38.24 -> 44.90 | Improved |
| add IFL | RIDE · ImageNet-GLT GLT protocol | accuracy: 43.00 -> 45.64 | Improved |
| add IFL | cRT · ImageNet-GLT ALT protocol | accuracy: 41.59 -> 45.47 | Improved |
| add IFL | RIDE · MSCOCO-GLT ALT protocol | accuracy: 58.90 -> 58.93 | No clear effect |
| add IFL | LDAM · ImageNet-GLT ALT protocol | accuracy: 42.66 -> 44.86 (BLSoftmax + IFL) | Improved |
| use vanilla center loss with a single environment (ablation) | ResNeXt-50 baseline · ImageNet-GLT GLT protocol | accuracy: 34.75 -> 32.82 | Hurt |
| use original IRM loss instead of IFL loss (ablation) | ResNeXt-50 · ImageNet-GLT GLT protocol | accuracy: 37.96 -> 36.03 | Hurt |
| increase environments from 2 to 3 (ablation) | IFL · ImageNet-GLT GLT protocol | accuracy: 37.96 -> 38.32 | No clear effect |
| replace CE with focal loss (ablation) | ResNeXt-50 baseline · ImageNet-GLT GLT protocol | accuracy: 34.75 -> 32.52 | Hurt |
Closest papers by meaning
- Deep Long-Tailed Learning: A Survey (TPAMI 2023 · similarity 0.90)
- Long-Tailed Learning as Multi-Objective Optimization (AAAI 2024 · similarity 0.89)
- Decoupling Representation and Classifier for Long-Tailed Recognition (ICLR 2020 · similarity 0.89)
- Feature Fusion from Head to Tail for Long-Tailed Visual Recognition (AAAI 2024 · similarity 0.88)
- Class and Attribute-Aware Logit Adjustment for Generalized Long-Tail Learning (AAAI 2025 · similarity 0.87)
- Long-Tailed Recognition via Information-Preservable Two-Stage Learning (NeurIPS 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.
