Research map / Label-efficient and robust learning / Zero-/few-shot and meta-learning
Meta Omnium: A Benchmark for General-Purpose Learning-To-Learn
Read the paperCheck 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 |
|---|---|---|---|
| Use ProtoNet rather than other evaluated learners | single-task meta-learning · Meta Omnium ID and OOD tasks | average rank: 1.5 average rank | Helped |
| Use multi-task rather than single-task meta-training | meta-learners evaluated on the same task families · Meta Omnium ID and OOD classification, segmentation, and keypoints | performance | Hurt |
| Use Proto-MAML instead of MAML | multi-task meta-learning · Meta Omnium multi-task tasks | average rank: 3.7 -> 2.3 | Helped |
| Use Meta-Curvature instead of MAML | single-task meta-learning · in-domain tasks | performance | Helped |
| Use Meta-Curvature instead of MAML | meta-learning · out-of-domain or multi-task tasks | performance | No clear effect |
| Use meta-learning rather than transfer learning | single-task and multi-task evaluation · Meta Omnium | performance | Helped |
| Use multi-task meta-learners for held-out regression tasks | models trained on classification, segmentation, and keypoints · four out-of-task regression datasets | performance | Mixed |
| Pre-train on ImageNet1k | multi-task ProtoNet · classification ID | accuracy (%): 66.0 -> 63.5 | Hurt |
| Pre-train on ImageNet1k | multi-task ProtoNet · segmentation ID | mIOU (%): 63.3 -> 62.0 | Hurt |
| Pre-train on ImageNet1k | multi-task ProtoNet · keypoints ID | PCK (%): 20.1 -> 20.1 | No clear effect |
| Pre-train on ImageNet1k | multi-task Proto-MAML · classification ID | accuracy (%): 58.5 -> 63.9 | Helped |
| Pre-train on ImageNet1k | multi-task Proto-MAML · classification OOD | accuracy (%): 63.7 -> 62.7 | Hurt |
Closest papers by meaning
- Data-Centric Meta-Learning for Robust Few-Shot Generalization (CVPR 2026 · similarity 0.89)
- What Matters for Meta-Learning Vision Regression Tasks? (CVPR 2022 · similarity 0.88)
- Unraveling Meta-Learning: Understanding Feature Representations for Few-Shot Tasks (ICML 2020 · similarity 0.88)
- A Dual Attention Network with Semantic Embedding for Few-Shot Learning (AAAI 2019 · similarity 0.87)
- Meta-Adapter: An Online Few-shot Learner for Vision-Language Model (NeurIPS 2023 · similarity 0.87)
- Unleashing the Power of Meta-tuning for Few-shot Generalization Through Sparse Interpolated Experts (ICML 2024 · similarity 0.86)
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.