Research map / Architectures and efficiency / Neural architecture search
MathNAS: If Blocks Have a Role in Mathematical Architecture Design
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 |
|---|---|---|---|
| search the MobileNetV3 space with MathNAS | MobileNetV3 · ImageNet-1K | Top-1 accuracy: 75.2 -> 75.9 | Helped |
| search the MobileNetV3 space with MathNAS | MobileNetV3 · ImageNet-1K | Top-1 accuracy: 76.3 -> 76.4 | Helped |
| search the MobileNetV3 space with MathNAS | MobileNetV3 · ImageNet-1K | Top-1 accuracy: 79.1 -> 79.2 | No clear effect |
| search with MathNAS rather than Transformer | SuperTransformer · WMT'14 En-De, Intel Xeon CPU | BLEU: 28.4 -> 28.8 | Helped |
| search with MathNAS rather than HAT | SuperTransformer · WMT'14 En-De, Nvidia TITAN Xp GPU | BLEU: 28.5 -> 28.6 | Helped |
| search using MathNAS | NAS-Bench-201 · CIFAR-10 test | Accuracy: 93.80±0.40 -> 93.31±0.00 | Hurt |
| search using MathNAS | NAS-Bench-201 · CIFAR-100 test | Accuracy: 71.53±1.51 -> 70.82±0.00 | Hurt |
| search using MathNAS | NAS-Bench-201 · ImageNet-16-120 test | Accuracy: 45.12±0.82 -> 46.53±0.00 | Helped |
| omit FLOPs information from accuracy prediction (ablation) | MathNAS predictor on MobileNetV3 · MobileNetV3 | RMSE: 0.117 -> 0.212 | Hurt |
| use a non-average-FLOPs baseline network (ablation) | MathNAS predictor on MobileNetV3 · MobileNetV3 | RMSE: 0.117 -> 0.513 | Hurt |
| use an arbitrary block instead of the first block as the baseline (ablation) | MathNAS predictor on MobileNetV3 · MobileNetV3 | RMSE: 0.117 -> 0.123 | Hurt |
| use a non-average-FLOPs baseline network (ablation) | MathNAS predictor on MobileNetV3 · MobileNetV3 | RMSE: 0.268 -> 0.408 | Hurt |
Closest papers by meaning
- Densely Connected Search Space for More Flexible Neural Architecture Search (CVPR 2020 · similarity 0.91)
- Neural Architecture Search on ImageNet in Four GPU Hours: A Theoretically Inspired Perspective (ICLR 2021 · similarity 0.90)
- Block-Wisely Supervised Neural Architecture Search With Knowledge Distillation (CVPR 2020 · similarity 0.90)
- AdvantageNAS: Efficient Neural Architecture Search with Credit Assignment (AAAI 2021 · similarity 0.90)
- Efficient Few-Shot Neural Architecture Search by Counting the Number of Nonlinear Functions (AAAI 2025 · similarity 0.89)
- HEP-NAS: Towards Efficient Few-shot Neural Architecture Search via Hierarchical Edge Partitioning (AAAI 2025 · 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.