Research map / Label-efficient and robust learning / Domain adaptation and generalization
Meta Learning on a Sequence of Imbalanced Domains With Difficulty Awareness
Zhenyi Wang, Tiehang Duan, Le Fang, Qiuling Suo, Mingchen Gao
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 M2D3 memory management (distribution and difficulty aware) in place of reservoir sampling (ablation) | PNet-RS · Imbalanced 6-domain meta-learning benchmark, 5-way 1-shot | accuracy (%): 34.68 -> 38.85 | Improved |
| add M2D3 memory management in place of reservoir sampling (ablation) | PNet-RS · Imbalanced 6-domain meta-learning benchmark, 5-way 5-shot | accuracy (%): 53.69 -> 57.95 | Improved |
| add adaptive memory task sampling (PETS) (ablation) | PNet with M2D3 · Imbalanced 6-domain meta-learning benchmark, 5-way 1-shot | accuracy (%): 38.85 -> 41.10 | Improved |
| add adaptive memory task sampling (PETS) (ablation) | PNet with M2D3 · Imbalanced 6-domain meta-learning benchmark, 5-way 5-shot | accuracy (%): 57.95 -> 60.37 | Improved |
| use proposed method instead of reservoir sampling | Prototypical Network · Imbalanced 6-domain meta-learning benchmark, 5-way 5-shot | accuracy (%): 53.69 -> 60.37 | Improved |
| use proposed method instead of reservoir sampling | ANIL · Imbalanced 6-domain meta-learning benchmark, 5-way 5-shot | accuracy (%): 48.72 -> 53.65 | Improved |
| use proposed method instead of reservoir sampling | ANIL · Imbalanced 6-domain meta-learning benchmark, 5-way 1-shot | accuracy (%): 32.11 -> 34.91 | Improved |
| increase memory capacity from 200 to 500 batches (ablation) | PNet-Ours · Imbalanced 6-domain meta-learning benchmark, 5-way 5-shot | accuracy (%): 59.86 -> 61.05 | Improved |
| increase memory capacity from 200 to 500 batches (ablation) | PNet-RS · Imbalanced 6-domain meta-learning benchmark, 5-way 5-shot | accuracy (%): 53.29 -> 55.95 | Improved |
| add continual learning replay methods (GSS, AGEM, MIR, MER) to PNet instead of RS | PNet-RS · Imbalanced 6-domain meta-learning benchmark, 5-way 1-shot | accuracy (%): 34.68 -> 34.07 (AGEM), 36.15 (GSS) | Mixed |
| replace sequential training with proposed method | PNet-Sequential · Imbalanced 6-domain meta-learning benchmark, 5-way 1-shot | accuracy (%): 31.82 -> 41.10 | Improved |
Closest papers by meaning
- Learning To Learn and Remember Super Long Multi-Domain Task Sequence (CVPR 2022 · similarity 0.88)
- Learning to Balance: Bayesian Meta-Learning for Imbalanced and Out-of-distribution Tasks (ICLR 2020 · similarity 0.87)
- Any-Way Meta Learning (AAAI 2024 · similarity 0.85)
- Addressing Imbalanced Domain-Incremental Learning through Dual-Balance Collaborative Experts (ICML 2025 · similarity 0.85)
- Fast Rate Bounds for Multi-Task and Meta-Learning with Different Sample Sizes (NeurIPS 2025 · similarity 0.85)
- Balanced Direction from Multifarious Choices: Arithmetic Meta-Learning for Domain Generalization (CVPR 2025 · similarity 0.85)
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.
