Research map / Architectures and efficiency / Neural architecture search
PINAT: A Permutation INvariance Augmented Transformer for NAS Predictor
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 |
|---|---|---|---|
| combine global self-attention features with partial permutation-invariant local features using PITE and PISA | PINAT · NAS-Bench-101 and NAS-Bench-201 | Kendall's Tau | Helped |
| add PIM to the predictor (ablation) | PINAT · NAS-Bench-101 | Kendall's Tau | Helped |
| add PITE while retaining PISA (ablation) | PINAT · NAS-Bench-101 | Kendall's Tau | Helped |
| add PISA while retaining PITE (ablation) | PINAT · NAS-Bench-101 | Kendall's Tau | Helped |
| use both PITE and PISA rather than either alone (ablation) | PINAT · NAS-Bench-101 | Kendall's Tau | Helped |
| use the residual, three-repeat PIM variant instead of the original PINE module (ablation) | PINAT · NAS-Bench-101 | Kendall's Tau | Helped |
| search architectures with PINAT rather than the strongest reported comparator | PINAT · CIFAR-10, DARTS search space | Best accuracy (%): 97.57 -> 97.58 | Helped |
| search architectures with PINAT-S rather than CTNAS | PINAT-S · ImageNet, ProxylessNAS search space | Top-1 (%): +0.5 | Helped |
| search architectures with PINAT rather than SGAS | PINAT · PPI | micro-F1 (%): 99.46 -> 99.48 | Helped |
| search architectures with PINAT rather than SGAS | PINAT · ModelNet40 | Overall accuracy (%): 92.93±0.19 -> 93.07 | Helped |
Closest papers by meaning
- TNASP: A Transformer-based NAS Predictor with a Self-evolution Framework (NeurIPS 2021 · similarity 0.93)
- Encodings for Prediction-based Neural Architecture Search (ICML 2024 · similarity 0.87)
- How Powerful are Performance Predictors in Neural Architecture Search? (NeurIPS 2021 · similarity 0.87)
- Surrogate NAS Benchmarks: Going Beyond the Limited Search Spaces of Tabular NAS Benchmarks (ICLR 2022 · similarity 0.86)
- ReNAS: Relativistic Evaluation of Neural Architecture Search (CVPR 2021 · similarity 0.86)
- Surprisingly Strong Performance Prediction with Neural Graph Features (ICML 2024 · similarity 0.85)
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.