mlsanityData to 7 Oct 2026

Prediction-powered Neural Architecture Search

arXiv 2610.01317 ↗First posted 1 Oct 2026cs.LGcited by 0

Not on the map. Why?

Abstract (arXiv)

Evaluating candidate architectures in neural architecture search (NAS) faces an inherent trade-off: on the one hand, reliable performance labels are limited because training and evaluating architectures is costly; on the other hand, zero-cost proxies (ZCPs) are cheap to compute at large scale but can be noisy. Yet, how to effectively combine these two sources of supervision remains unclear. In this paper, we propose PPNAS, a novel prediction-powered inference (PPI) approach for NAS. PPNAS fuses (1) a small set of architectures with observed performance labels and (2) a large set of architectures with ZCP information. To combine these two sources of supervision, PPNAS exploits the ordinal information provided by ZCPs to construct additional pairwise ranking supervision, while PPI debiases systematic discrepancies between ZCP-based and true performance rankings. We evaluate PPNAS in end-to-end predictor-based NAS, where it achieves state-of-the-art under limited evaluation budgets. To the best of our knowledge, PPNAS is the first prediction-powered approach for label-efficient NAS.

Key facts. No citing papers in the map yet (data through 7 Oct 2026). Builds on 34 of its 41 references that are in the map. First posted on arXiv 1 Oct 2026.

Builds on

34 of 41 references.

References · 34

  1. 2018DARTS: Differentiable Architecture Searchcited by 1,782
  2. 2016Neural Architecture Search with Reinforcement Learningcited by 1,489
  3. 2018Regularized Evolution for Image Classifier Architecture Searchcited by 902
  4. 2018Efficient Neural Architecture Search via Parameter Sharingcited by 831
  5. 2018SNIP: Single-shot Network Pruning based on Connection Sensitivitycited by 802
  6. 2020Pruning neural networks without any data by iteratively conserving synaptic flowcited by 475
  7. 2020Picking Winning Tickets Before Training by Preserving Gradient Flowcited by 467
  8. 2020NAS-Bench-201: Extending the Scope of Reproducible Neural Architecture Searchcited by 412
  9. 2017A Downsampled Variant of ImageNet as an Alternative to the CIFAR datasetscited by 374
  10. 2019NAS-Bench-101: Towards Reproducible Neural Architecture Searchcited by 340
  11. 2020Neural Architecture Search without Trainingcited by 241
  12. 2023Prediction-Powered Inferencecited by 214
22 more
  1. 2021Zero-Cost Proxies for Lightweight NAScited by 194
  2. 2019BANANAS: Bayesian Optimization with Neural Architectures for Neural Architecture Searchcited by 175
  3. 2023Neural Architecture Search: Insights from 1000 Paperscited by 136
  4. 2020BRP-NAS: Prediction-based NAS using GCNscited by 109
  5. 2020NATS-Bench: Benchmarking NAS Algorithms for Architecture Topology and Sizecited by 98
  6. 2021Zen-NAS: A Zero-Shot NAS for High-Performance Deep Image Recognitioncited by 71
  7. 2020Does Unsupervised Architecture Representation Learning Help Neural Architecture Search?cited by 58
  8. 2021TransNAS-Bench-101: Improving Transferability and Generalizability of Cross-Task Neural Architecture Searchcited by 57
  9. 2023ZiCo: Zero-shot NAS via Inverse Coefficient of Variation on Gradientscited by 57
  10. 2023Cross-Prediction-Powered Inferencecited by 57
  11. 2020Evaluating Efficient Performance Estimators of Neural Architecturescited by 54
  12. 2024AutoEval Done Right: Using Synthetic Data for Model Evaluationcited by 53
  13. 2020Semi-Supervised Neural Architecture Searchcited by 45
  14. 2020A Study on Encodings for Neural Architecture Searchcited by 37
  15. 2022NAS-Bench-Suite-Zero: Accelerating Research on Zero Cost Proxiescited by 36
  16. 2019ReNAS:Relativistic Evaluation of Neural Architecture Searchcited by 35
  17. 2024Prediction-Powered Ranking of Large Language Modelscited by 29
  18. 2021ProxyBO: Accelerating Neural Architecture Search via Bayesian Optimization with Zero-cost Proxiescited by 20
  19. 2021RankNAS: Efficient Neural Architecture Search by Pairwise Rankingcited by 12
  20. 2024Surprisingly Strong Performance Prediction with Neural Graph Featurescited by 11
  21. 2026Nonparametric LLM Evaluation from Preference Datacited by 6
  22. 2025Loss Functions for Predictor-based Neural Architecture Searchcited by 1

Built on by

Not cited yet.