Track trials, findings, and clinical evidence as it lands.
Search papers, labs, and topics across Lattice.
This paper addresses the high evaluation cost in hardware-aware Neural Architecture Search (NAS) by adapting conformal prediction for online, RL-driven settings. It introduces online conformal calibration techniques, including Adaptive Conformal Inference variants, to restore steerable coverage guarantees violated by the non-exchangeability of RL-generated candidates. The method achieves precise coverage control across diverse NAS tasks, pruning a significant portion of evaluations without accuracy degradation, outperforming static calibration and Gaussian-process baselines.
Steerable conformal coverage guarantees are now achievable in RL-driven NAS, enabling precise pruning of millions of architecture evaluations without sacrificing accuracy.
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 https://github.com/Vicomtech/rl-hw-nas.