Research map / Architectures and efficiency / Neural architecture search
Prioritized Architecture Sampling With Monto-Carlo Tree Search
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 a dependency-aware MCTS tree to sample architectures instead of independent layer sampling | One-shot NAS on NAS-Bench-Macro · NAS-Bench-Macro | Spearman rho: 88.96 -> 90.63 | Helped |
| Add uniform-sampling warm-up before MCTS sampling (ablation) | MCT-NAS · NAS-Bench-Macro | Spearman rho: 90.63 -> 91.87 | Helped |
| Replace evolutionary search with MCTS search and hierarchical updates (ablation) | MCT-NAS on ImageNet · ImageNet, 330M FLOPs | Top-1 accuracy: 75.94% -> 76.44% | Helped |
| Add FLOPs reduction during MCT training (ablation) | MCT-NAS on ImageNet · ImageNet, 330M FLOPs | Top-1 accuracy: 76.44% -> 76.49% | Helped |
| Update the MCT during training using subnet losses (ablation) | MCT-NAS on ImageNet · ImageNet, 330M FLOPs | Top-1 accuracy: 76.44% -> 76.54% | Helped |
| Combine FLOPs reduction with MCT training updates (ablation) | MCT-NAS on ImageNet · ImageNet, 330M FLOPs | Top-1 accuracy: 76.44% -> 76.65% | Helped |
| Add UCT-based sampling during training (ablation) | MCT-NAS on ImageNet · ImageNet, 330M FLOPs | Top-1 accuracy: 76.21% -> 76.35% | Helped |
| Add node communication to share operation rewards across matching nodes (ablation) | MCT-NAS on ImageNet · ImageNet, 330M FLOPs | Top-1 accuracy: 76.35% -> 76.62% | Helped |
| Add hierarchical node updates using validation data (ablation) | MCT-NAS on ImageNet · ImageNet, 330M FLOPs | Top-1 accuracy: 76.62% -> 76.94% | Helped |
| Increase the hierarchical selection visit threshold (ablation) | MCT-NAS · ImageNet, 330M FLOPs | validation accuracy | Mixed |
| Increase the number of searched architectures (ablation) | MCT-NAS · NAS-Bench-Macro | top accuracy: 50 | Helped |
| Increase supernet training from 60 to 120 epochs (ablation) | MCT-NAS · ImageNet, 330M FLOPs | Top-1 accuracy: 60; 120 | No clear effect |
Closest papers by meaning
- Neural Architecture Search Using Deep Neural Networks and Monte Carlo Tree Search (AAAI 2020 · similarity 0.92)
- BayesNAS: A Bayesian Approach for Neural Architecture Search (ICML 2019 · similarity 0.90)
- Bridging the Gap between Sample-based and One-shot Neural Architecture Search with BONAS (NeurIPS 2020 · similarity 0.90)
- HEP-NAS: Towards Efficient Few-shot Neural Architecture Search via Hierarchical Edge Partitioning (AAAI 2025 · similarity 0.88)
- Efficient Few-Shot Neural Architecture Search by Counting the Number of Nonlinear Functions (AAAI 2025 · similarity 0.88)
- MOTE-NAS: Multi-Objective Training-based Estimate for Efficient Neural Architecture Search (NeurIPS 2024 · 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.