Research map / Architectures and efficiency / Neural architecture search
Neural Architecture Search With Representation Mutual Information
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 |
|---|---|---|---|
| Use RMI-NAS instead of RL search | NAS-Bench-201 architectures · CIFAR-10 test | accuracy (%): 93.85 ± 0.37 -> 94.28 ± 0.10 | Helped |
| Use RMI-NAS instead of RL search | NAS-Bench-201 architectures · CIFAR-100 test | accuracy (%): 71.71 ± 1.09 -> 73.36 ± 0.19 | Helped |
| Use RMI-NAS instead of RL search | NAS-Bench-201 architectures · ImageNet16-120 test | accuracy (%): 45.24 ± 1.18 -> 46.34 ± 0.00 | Helped |
| Use RMI-NAS instead of RL search | NAS-Bench-201 search · NAS-Bench-201 | search cost (seconds): 27870.7 -> 1258.21 | Helped |
| Use RMI-NAS instead of P-DARTS | DARTS search space · CIFAR-10 retrain | test error (%): 2.70 ± 0.15* -> 2.64 ± 0.04 | Helped |
| Increase the RMI-loss weight β from classification-only to β = 0.8 (ablation) | RMI performance estimation · NAS-Bench-201 CIFAR-10 | Kendall's τ | Helped |
| Use an accurate teacher architecture α+ with accuracy greater than 85% (ablation) | RMI performance estimation · NAS-Bench-201 | Kendall's τ | Helped |
| Increase the optimization iterations used for RMI estimation (ablation) | RMI performance estimation · NAS-Bench-201 | Kendall's τ | Helped |
| Replace standard training-based performance estimation with RMI | RL, REA, and random search · NAS-Bench-201 | search speed: 18× Faster; 19.2× Faster; 19× Faster | Helped |
Closest papers by meaning
- Evaluating The Search Phase of Neural Architecture Search (ICLR 2020 · similarity 0.92)
- Neural Architecture Generator Optimization (NeurIPS 2020 · similarity 0.91)
- NASI: Label- and Data-agnostic Neural Architecture Search at Initialization (ICLR 2022 · similarity 0.91)
- Speedy Performance Estimation for Neural Architecture Search (NeurIPS 2021 · similarity 0.91)
- einspace: Searching for Neural Architectures from Fundamental Operations (NeurIPS 2024 · similarity 0.90)
- Insights from the Use of Previously Unseen Neural Architecture Search Datasets (CVPR 2024 · similarity 0.90)
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.