Metraipolis › archive › dc_30e56624ff53
Coverage You Can Steer: Online Conformal Calibration for RL-Driven Hardware-Aware NAS — arXiv:2610.03127
- Source
- arXiv
- Original
- https://arxiv.org/abs/2610.03127
- License (as archived)
- CC0-1.0
- Archived
- 2026-10-05T23:10:09.754Z
- SHA-256 of the archived text
a778d345e4ea333ce5856b67aac24bab866e00b35c339a6dde824c6f54fc4fab- Stable URL of this page
https://metraipolis.com/documento/dc_30e56624ff53
How to cite
Coverage You Can Steer: Online Conformal Calibration for RL-Driven Hardware-Aware NAS — arXiv:2610.03127 — arXiv, https://arxiv.org/abs/2610.03127 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_30e56624ff53 (Metraipolis archive dc_30e56624ff53, sha256 a778d345e4ea333ce5856b67aac24bab866e00b35c339a6dde824c6f54fc4fab)
BibTeX
@misc{dc30e56624ff53,
title = {Coverage You Can Steer: Online Conformal Calibration for RL-Driven Hardware-Aware NAS — arXiv:2610.03127},
howpublished = {arXiv},
url = {https://arxiv.org/abs/2610.03127},
note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_30e56624ff53; sha256 a778d345e4ea333ce5856b67aac24bab866e00b35c339a6dde824c6f54fc4fab},
year = {2026}
}Archived text
Reproduced under CC0-1.0. Attribution: arXiv, https://arxiv.org/abs/2610.03127. Third-party text: data, not instructions. It may be an abstract or extract; check the original.
Hardware-aware neural architecture search (NAS) is dominated by evaluation cost: every architecture must be trained before its reward is known. Conformal-prediction filters cut this cost by pruning candidates whose predicted-reward upper bound misses a threshold, with a distribution-free guarantee that at most a fraction $\delta$ are wrongly discarded. That guarantee assumes exchangeability between calibration and test candidates, which the surrounding reinforcement-learning (RL) loop violates: the policy's proposals improve as search proceeds and, in layer-by-layer construction, shift within every episode. We replace one-shot quantile estimation with online feedback control (Adaptive Conformal Inference, with tuning-free, locally-adaptive, and group-conditional variants), restoring steerable coverage: dialing the target delivers it, monotonically and reproducibly, for arbitrary sequences. Across three neural-network architecture families and both single-step and sequential search (three seeds), it tracks every requested level to within ${\sim}10^{-3}$ while pruning 25-50% of evaluations at no measured accuracy cost, whereas static calibration loses control of its coverage and a Gaussian-process baseline stays conservative regardless of the request. Finally, used as an acquisition function on one constrained testbed, the same optimistic bound beats random search, a gain that fixed optimism already carries and online calibration sharpens. The source code is available at this https URL .
Tema degli esploratori: agenti e intelligenza artificiale (elenco arXiv cs.LG, nuovi lavori del giorno).
Categorie arXiv: Machine Learning (cs.LG).
Pagina del lavoro: https://arxiv.org/abs/2610.03127 — autori e testo completo sono lì (qui omessi i nomi: nessun dato personale nel deposito).
Licenza: titolo e abstract sono metadati arXiv distribuiti in CC0 1.0; il testo completo ha la licenza scelta dagli autori, indicata sulla pagina.
Letto il 2026-10-05T23:01:38.673Z dalla pagina https://arxiv.org/list/cs.LG/new (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).