Research map / Architectures and efficiency / Neural architecture search
TNASP: A Transformer-based NAS Predictor with a Self-evolution Framework
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 |
|---|---|---|---|
| add self-evolution framework | TNASP · NAS-Bench-101, 424 train (all test) | Kendall's Tau: 0.705 -> 0.722 | Helped |
| add self-evolution framework | TNASP · NAS-Bench-101, 4236 train | Kendall's Tau: 0.820 -> 0.820 | No clear effect |
| add self-evolution framework | TNASP · NAS-Bench-201, 78 train | Kendall's Tau: 0.539 -> 0.565 | Helped |
| add self-evolution framework | TNASP · NAS-Bench-201, 1563 train | Kendall's Tau: 0.724 -> 0.726 | No clear effect |
| add self-evolution framework | Neural Predictor · NAS-Bench-101, 100 train | Kendall's Tau: 0.391 -> 0.458 | Helped |
| add self-evolution framework | NAO · NAS-Bench-201, 78 train | Kendall's Tau: 0.467 -> 0.511 | Helped |
| use Laplacian matrix positional encoding instead of no informative encoding (random) (ablation) | TNASP · NAS-Bench-101, 4236 train | Kendall's Tau: 0.365 -> 0.820 | Helped |
| use Laplacian instead of Adjacency positional encoding (ablation) | TNASP · NAS-Bench-101, 100 train | Kendall's Tau: 0.620 -> 0.600 | Hurt |
| use Laplacian instead of Adjacency positional encoding (ablation) | TNASP · NAS-Bench-101, 4236 train | Kendall's Tau: 0.802 -> 0.820 | Helped |
| use unnormalized instead of normalized Laplacian (ablation) | TNASP · NAS-Bench-101, 4236 train | Kendall's Tau: 0.813 -> 0.820 | Helped |
| use Laplacian matrix linear projection instead of Laplacian eigenvectors (ablation) | TNASP · NAS-Bench-101, 4236 train | Kendall's Tau: 0.656 -> 0.820 | Helped |
| increase validation evaluations in SE from 1 to 10 (ablation) | TNASP + SE · NAS-Bench-101 | Kendall's Tau | Helped |
Closest papers by meaning
- PINAT: A Permutation INvariance Augmented Transformer for NAS Predictor (AAAI 2023 · similarity 0.93)
- How Powerful are Performance Predictors in Neural Architecture Search? (NeurIPS 2021 · similarity 0.89)
- Encodings for Prediction-based Neural Architecture Search (ICML 2024 · similarity 0.88)
- BRP-NAS: Prediction-based NAS using GCNs (NeurIPS 2020 · similarity 0.87)
- NAS-Bench-201: Extending the Scope of Reproducible Neural Architecture Search (ICLR 2020 · similarity 0.87)
- Surprisingly Strong Performance Prediction with Neural Graph Features (ICML 2024 · similarity 0.86)
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.