Research map / Architectures and efficiency / RNNs and normalization
How to train your MAML
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 |
|---|---|---|---|
| Combine all proposed fixes | MAML · Mini-ImageNet 5-way 1-shot | accuracy: 48.25$\mypm$0.62\% -> 52.15$\mypm$0.26\% | Helped |
| Combine all proposed fixes | MAML · Mini-ImageNet 5-way 5-shot | accuracy: 64.39$\mypm$0.31\% -> 68.32$\mypm$0.44\% | Helped |
| Add per-layer, per-step learning rates and gradient directions (ablation) | MAML · Omniglot 20-way 1-shot | accuracy: 91.27$\mypm$1.07\% -> 95.77$\mypm$0.38\% | Helped |
| Add per-step batch-normalization weights, biases, and running statistics (ablation) | MAML · Omniglot 20-way 1-shot | accuracy: 91.27$\mypm$1.07\% -> 95.35$\mypm$0.23\% | Helped |
| Add multi-step loss optimization (ablation) | MAML · Omniglot 20-way 1-shot | accuracy: 91.27$\mypm$1.07\% -> 91.53$\mypm$0.69\% | No clear effect |
| Add cosine annealing of the meta-optimizer learning rate (ablation) | MAML · Omniglot 20-way 1-shot | accuracy: 91.27$\mypm$1.07\% -> 93.03$\mypm$0.44\% | Helped |
| Add derivative-order annealing (ablation) | MAML · Omniglot 20-way 1-shot | accuracy: 91.27$\mypm$1.07\% -> 92.3$\mypm$0.55\% | Helped |
| Use 5 rather than 1 inner step (ablation) | MAML++ · Mini-ImageNet 5-way 1-shot | accuracy: 51.05$\mypm$0.31\% -> 52.15$\mypm$0.26\% | Helped |
| Use MAML++ instead of MAML | Mini-ImageNet training · Mini-ImageNet | milliseconds per iteration: 294.373 -> 275.319 | Helped |
| Use MAML++ with 1 inner step | MAML · Omniglot 5-way 5-shot | accuracy: 99.9$\mypm$0.1\% -> 99.93\% | No clear effect |
Closest papers by meaning
- How to Train Your MAML to Excel in Few-Shot Classification (ICLR 2022 · similarity 0.93)
- Repurposing Pretrained Models for Robust Out-of-domain Few-Shot Learning (ICLR 2021 · similarity 0.91)
- Global Convergence of MAML and Theory-Inspired Neural Architecture Search for Few-Shot Learning (CVPR 2022 · similarity 0.90)
- Sharp-MAML: Sharpness-Aware Model-Agnostic Meta Learning (ICML 2022 · similarity 0.90)
- Enabling Few-Shot Learning with PID Control: A Layer Adaptive Optimizer (ICML 2024 · similarity 0.88)
- Hypernetwork Approach to Bayesian MAML (Student Abstract) (AAAI 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.