Research map / Optimization / Optimizers
Binomial Gradient-Based Meta-Learning for Enhanced Meta-Gradient Estimation
Yilang Zhang, Abraham Jaeger Mountain, Bingcong Li, Georgios B. Giannakis
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 |
|---|---|---|---|
| use binomial expansion instead of truncated backprop (L=1) | TruncMAML · miniImageNet 1-shot | accuracy (%): 44.53 -> 45.50 | Improved |
| use binomial expansion instead of truncated backprop (L=2) | TruncMAML · miniImageNet 1-shot | accuracy (%): 44.93 -> 46.23 | Improved |
| use binomial expansion instead of truncated backprop (L=3) | TruncMAML · miniImageNet 1-shot | accuracy (%): 44.33 -> 46.00 | Improved |
| use binomial expansion instead of truncated backprop (L=1) | TruncMAML · tieredImageNet 1-shot | accuracy (%): 43.67 -> 46.23 | Improved |
| use binomial expansion instead of truncated backprop (L=1 and 2) | TruncMAML · miniImageNet 5-shot | accuracy (%): 62.43 -> 62.36 (L=1), 63.61 -> 63.49 (L=2) | Mixed |
| use binomial expansion instead of truncated backprop, averaged over L | TruncMAML · 1-shot vs 5-shot | average accuracy gap: +1.33 average (1-shot), +0.27 average (5-shot) | Mixed |
| use BinomMAML instead of iMAML (L=4) | iMAML · miniImageNet 1-shot | accuracy (%): 45.43 -> 46.24 | Improved |
| use BinomMAML instead of iMAML (L=4) | iMAML · tieredImageNet 5-shot | accuracy (%): 61.80 -> 64.63 | Improved |
| increase truncation L from 1 to 4 for BinomMAML (ablation) | BinomMAML · miniImageNet 1-shot | accuracy (%): 45.50 -> 46.24 | Improved |
| increase truncation L from 1 to 4 for TruncMAML (ablation) | TruncMAML · miniImageNet 1-shot | accuracy (%): 44.53 -> 44.83 | No clear effect |
| increase truncation L from 1 to 4 for BinomMAML (ablation) | BinomMAML · tieredImageNet 5-shot | accuracy (%): 64.49 -> 64.63 | No clear effect |
| BinomMAML versus truncated backprop in meta-gradient error (L=4) | TruncMAML · synthetic sinusoid regression | meta-gradient error | Improved |
Closest papers by meaning
- Recasting Gradient-Based Meta-Learning as Hierarchical Bayes (ICLR 2018 · similarity 0.88)
- On the Stability and Generalization of Meta-Learning: the Impact of Inner-Levels (NeurIPS 2025 · similarity 0.88)
- MetaDiff: Meta-Learning with Conditional Diffusion for Few-Shot Learning (AAAI 2024 · similarity 0.88)
- Efficient Meta Learning via Minibatch Proximal Update (NeurIPS 2019 · similarity 0.88)
- Provable Generalization of Overparameterized Meta-learning Trained with SGD (NeurIPS 2022 · similarity 0.87)
- Memory-Reduced Meta-Learning with Guaranteed Convergence (AAAI 2025 · similarity 0.87)
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.
