Research map / Label-efficient and robust learning / Zero-/few-shot and meta-learning
Meta-Learning Requires Meta-Augmentation
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 |
|---|---|---|---|
| add label shuffling across tasks (intershuffle) to make tasks mutually exclusive (ablation) | MAML · Mini-ImageNet 1-shot 5-way | accuracy (%): 30.2% -> 46.0% | Helped |
| shuffle label order within fixed class groups (intrashuffle) vs none (ablation) | MAML · Mini-ImageNet 1-shot 5-way | accuracy (%): 30.2% -> 42.7% | Helped |
| resample classes across tasks (intershuffle vs intrashuffle) (ablation) | MAML · Mini-ImageNet 1-shot 5-way | accuracy (%): 42.7% -> 46.0% | Helped |
| add intershuffle label augmentation (ablation) | Prototypical Networks · Mini-ImageNet 1-shot 5-way | accuracy (%): 32.5% -> 37.2% | Helped |
| add intershuffle label augmentation (ablation) | Matching Networks · Mini-ImageNet 1-shot 5-way | accuracy (%): 33.8% -> 39.8% | Helped |
| add intershuffle label augmentation (ablation) | MAML · Omniglot 1-shot 5-way | accuracy (%): 98.1% -> 98.7% | No clear effect |
| add intershuffle label augmentation (ablation) | MAML · D'Claw 1-shot 2-way | accuracy (%): 72.5% -> 83.1% | Helped |
| add discrete-noise label augmentation to regression targets (ablation) | MR-MAML / CNP on Pascal3D pose regression · Pascal3D pose regression | MSE | Helped |
| add continuous uniform noise instead of discrete noise (ablation) | MR-MAML · Pascal3D pose regression | MSE | Hurt |
| remove weight decay from MAML baseline | MAML · Pascal3D pose regression | MSE | Helped |
| increase number of discrete noise values beyond 4 (MR-MAML) or 2 (MAML) (ablation) | meta-augmentation on Pascal3D · Pascal3D pose regression | MSE | Hurt |
| add label noise augmentation to MAML on sinusoid regression (ablation) | MAML · Sinusoid regression | MSE | Helped |
Closest papers by meaning
- Improving Generalization of Meta-Learning With Inverted Regularization at Inner-Level (CVPR 2023 · similarity 0.92)
- Improving Generalization in Meta-learning via Task Augmentation (ICML 2021 · similarity 0.91)
- More Flexible PAC-Bayesian Meta-Learning by Learning Learning Algorithms (ICML 2024 · similarity 0.88)
- Learning to Learn with Contrastive Meta-Objective (NeurIPS 2025 · similarity 0.88)
- An Information-Theoretic Analysis of In-Context Learning (ICML 2024 · similarity 0.88)
- Provable Meta-Learning with Low-Rank Adaptations (NeurIPS 2025 · similarity 0.88)
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.