Research map / Architectures and efficiency / Neural architecture search
Learning To Learn by Jointly Optimizing Neural Architecture and Weights
Check 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 |
|---|---|---|---|
| jointly optimize architecture connections and meta-weights with CAML | CAML · Mini-Imagenet 5-way 5-shot | accuracy: 64.1 ± 1.1 -> 68.1 ± 0.3 | Helped |
| jointly optimize architecture connections and meta-weights with CAML | CAML · FC100 5-way 5-shot | accuracy: 52.2 ± 1.2 -> 53.6 ± 0.2 | Helped |
| use CAML rather than OPML (ablation) | Mini-Imagenet meta-learning · Mini-Imagenet 5-way 5-shot, PCC and kept weights | accuracy: 64.2 ± 0.2 -> 68.1 ± 0.3 | Helped |
| add progressive connection consolidation during CAML search (ablation) | CAML · Mini-Imagenet 5-way 5-shot, train from scratch | accuracy: 62.6 ± 0.1 -> 67.4 ± 0.5 | Helped |
| add progressive connection consolidation and retain searched weights (ablation) | CAML · Mini-Imagenet 5-way 5-shot | accuracy: 62.8 ± 0.4 -> 68.1 ± 0.3 | Helped |
| retain searched meta-weights instead of using random initialization (ablation) | CAML with PCC · Mini-Imagenet 5-way 5-shot | accuracy: 67.4 ± 0.5 -> 68.1 ± 0.3 | Helped |
| use layer-confidence-based pruning rather than forward-order pruning (ablation) | CAML with PCC · Mini-Imagenet 5-way 5-shot, searched weights | accuracy: 66.0 ± 0.1 -> 68.1 ± 0.2 | Helped |
| use layer-confidence-based pruning rather than backward-order pruning (ablation) | CAML with PCC · Mini-Imagenet 5-way 5-shot, searched weights | accuracy: 64.5 ± 0.2 -> 68.1 ± 0.2 | Helped |
| use layer-confidence-based pruning rather than entropy-based pruning (ablation) | CAML with PCC · Mini-Imagenet 5-way 5-shot, searched weights | accuracy: 67.7 ± 0.1 -> 68.1 ± 0.2 | Helped |
| use layer-confidence-based pruning rather than variance-based pruning (ablation) | CAML with PCC · Mini-Imagenet 5-way 5-shot, searched weights | accuracy: 67.8 ± 0.1 -> 68.1 ± 0.2 | Helped |
| use the S1 operation search space (ablation) | CAML · Mini-Imagenet 5-way 5-shot | accuracy: 62.1± 0.9 -> 65.4± 0.1 | Helped |
| use Reptile instead of MAML as the meta-learning method (ablation) | CAML · Mini-Imagenet 5-way 5-shot | accuracy: 68.1± 0.3 -> 68.5± 0.3 | Helped |
Closest papers by meaning
- Towards Fast Adaptation of Neural Architectures with Meta Learning (ICLR 2020 · similarity 0.91)
- Memory-Reduced Meta-Learning with Guaranteed Convergence (AAAI 2025 · similarity 0.90)
- Meta-Learning of Neural Architectures for Few-Shot Learning (CVPR 2020 · similarity 0.90)
- Learning Large-scale Neural Fields via Context Pruned Meta-Learning (NeurIPS 2023 · similarity 0.90)
- Bridging Multi-Task Learning and Meta-Learning: Towards Efficient Training and Effective Adaptation (ICML 2021 · similarity 0.89)
- Online Meta-Learning via Learning with Layer-Distributed Memory (NeurIPS 2021 · similarity 0.89)
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.