SEval-NAS: A Search-Agnostic Evaluation for Neural Architecture Search

TL;DR

SEval-NAS introduces a flexible, search-agnostic evaluation method for neural architecture search that predicts performance metrics like accuracy, latency, and memory, enhancing hardware-aware NAS without hardcoded evaluation procedures.

Contribution

It proposes a novel metric-evaluation mechanism converting architectures into vectors for performance prediction, enabling flexible, hardware-aware NAS evaluation.

Findings

SEval-NAS predicts latency and memory more accurately than accuracy.

The method effectively ranks architectures generated by FreeREA.

SEval-NAS maintains search efficiency with minimal algorithmic changes.

Abstract

Neural architecture search (NAS) automates the discovery of neural networks that meet specified criteria, yet its evaluation procedures are often hardcoded, limiting the ability to introduce new metrics. This issue is especially pronounced in hardware-aware NAS, where objectives depend on target devices such as edge hardware. To address this limitation, we propose SEval-NAS, a metric-evaluation mechanism that converts architectures to strings, embeds them as vectors, and predicts performance metrics. Using NATS-Bench and HW-NAS-Bench, we evaluated accuracy, latency, and memory. Kendall's $\tau$ correlations showed stronger latency and memory predictions than accuracy, indicating the suitability of SEval-NAS as a hardware cost predictor. We further integrated SEval-NAS into FreeREA to evaluate metrics not originally included. The method successfully ranked FreeREA-generated architectures, maintained search time, and required minimal algorithmic changes. Our implementation is available at: https://github.com/Analytics-Everywhere-Lab/neural-architecture-search