FOX-NAS: Fast, On-device and Explainable Neural Architecture Search

TL;DR

FOX-NAS introduces a rapid, explainable neural architecture search method using simulated annealing and multivariate regression, enabling efficient edge deployment and outperforming some existing models in latency and accuracy.

Contribution

The paper presents FOX-NAS, a novel, fast, and explainable NAS approach that reduces predictor training time and enhances edge deployment efficiency.

Findings

FOX-NAS models outperform some popular architectures in latency and accuracy.

FOX-NAS matches MobileNetV2 and EfficientNet-Lite0 accuracy with significantly less latency.

FOX-NAS won 3rd place in the 2020 Low-Power Computer Vision Challenge.

Abstract

Neural architecture search can discover neural networks with good performance, and One-Shot approaches are prevalent. One-Shot approaches typically require a supernet with weight sharing and predictors that predict the performance of architecture. However, the previous methods take much time to generate performance predictors thus are inefficient. To this end, we propose FOX-NAS that consists of fast and explainable predictors based on simulated annealing and multivariate regression. Our method is quantization-friendly and can be efficiently deployed to the edge. The experiments on different hardware show that FOX-NAS models outperform some other popular neural network architectures. For example, FOX-NAS matches MobileNetV2 and EfficientNet-Lite0 accuracy with 240% and 40% less latency on the edge CPU. FOX-NAS is the 3rd place winner of the 2020 Low-Power Computer Vision Challenge (LPCVC), DSP classification track. See all evaluation results at https://lpcv.ai/competitions/2020. Search code and pre-trained models are released at https://github.com/great8nctu/FOX-NAS.