Research map / Architectures and efficiency / RNNs and normalization
Meta-Learning with Implicit Gradients
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 Hessian-free optimization instead of gradient descent in the inner loop (ablation) | iMAML · Omniglot 5-way 1-shot | accuracy: 99.16 $\pm$ 0.35\% -> \textbf{99.50 $\pm$ 0.26\%} | Helped |
| Use Hessian-free optimization instead of gradient descent in the inner loop (ablation) | iMAML · Omniglot 5-way 5-shot | accuracy: 99.67 $\pm$ 0.12\% -> 99.74 $\pm$ 0.11\% | Helped |
| Use Hessian-free optimization instead of gradient descent in the inner loop (ablation) | iMAML · Omniglot 20-way 1-shot | accuracy: 94.46 $\pm$ 0.42\% -> \textbf{96.18 $\pm$ 0.36\%} | Helped |
| Use Hessian-free optimization instead of gradient descent in the inner loop (ablation) | iMAML · Omniglot 20-way 5-shot | accuracy: 98.69 $\pm$ 0.1\% -> \textbf{99.14 $\pm$ 0.1\%} | Helped |
| Use Hessian-free optimization instead of gradient descent in the inner loop (ablation) | iMAML · Mini-ImageNet 5-way 1-shot | accuracy: 48.96 $\pm$ 1.84 \% -> 49.30 $\pm$ 1.88 \% | Mixed |
| Use implicit gradients with gradient descent inner optimization instead of MAML's path backpropagation | iMAML · Omniglot 5-way 1-shot | accuracy: 98.7 $\pm$ 0.4\% -> 99.16 $\pm$ 0.35\% | Mixed |
| Use implicit gradients with gradient descent inner optimization instead of MAML's path backpropagation | iMAML · Mini-ImageNet 5-way 1-shot | accuracy: 48.70 $\pm$ 1.84 \% -> 48.96 $\pm$ 1.84 \% | Mixed |
Closest papers by meaning
- MetaDiff: Meta-Learning with Conditional Diffusion for Few-Shot Learning (AAAI 2024 · similarity 0.90)
- Rapid Learning or Feature Reuse? Towards Understanding the Effectiveness of MAML (ICLR 2020 · similarity 0.90)
- Meta-Learning with Neural Tangent Kernels (ICLR 2021 · similarity 0.89)
- Scalable Bayesian Meta-Learning through Generalized Implicit Gradients (AAAI 2023 · similarity 0.89)
- A Nested Bi-level Optimization Framework for Robust Few Shot Learning (AAAI 2022 · similarity 0.89)
- Memory-Reduced Meta-Learning with Guaranteed Convergence (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.