AOWS: Adaptive and optimal network width search with latency constraints

TL;DR

This paper presents AOWS, a novel one-shot neural architecture search method that efficiently finds optimal network widths under latency constraints by modeling latency, scoring configurations with MRF, and adaptively sampling channels.

Contribution

AOWS introduces a black-box latency estimation, an MRF-based scoring system, and an adaptive sampling scheme for efficient, constrained neural architecture search.

Findings

Achieves better accuracy than state-of-the-art efficient networks.

Successfully adapts network widths to various hardware platforms.

Maintains latency constraints while improving performance.

Abstract

Neural architecture search (NAS) approaches aim at automatically finding novel CNN architectures that fit computational constraints while maintaining a good performance on the target platform. We introduce a novel efficient one-shot NAS approach to optimally search for channel numbers, given latency constraints on a specific hardware. We first show that we can use a black-box approach to estimate a realistic latency model for a specific inference platform, without the need for low-level access to the inference computation. Then, we design a pairwise MRF to score any channel configuration and use dynamic programming to efficiently decode the best performing configuration, yielding an optimal solution for the network width search. Finally, we propose an adaptive channel configuration sampling scheme to gradually specialize the training phase to the target computational constraints. Experiments on ImageNet classification show that our approach can find networks fitting the resource constraints on different target platforms while improving accuracy over the state-of-the-art efficient networks.