Neural Architecture Search using Property Guided Synthesis

TL;DR

This paper introduces a novel neural architecture search method, alphaNAS, that searches efficiently in a reduced abstract space of program properties, leading to models with fewer parameters and FLOPS while maintaining or improving accuracy.

Contribution

The paper presents a property-guided synthesis approach for NAS that operates in an abstract space, enabling efficient exploration of larger design spaces with minimal manual effort.

Findings

Achieved models with 96% fewer parameters on CIFAR-10.

Improved model efficiency on ImageNet with no accuracy loss.

Generated models outperform or match state-of-the-art in FLOPS and parameters.

Abstract

In the past few years, neural architecture search (NAS) has become an increasingly important tool within the deep learning community. Despite the many recent successes of NAS, however, most existing approaches operate within highly structured design spaces, and hence explore only a small fraction of the full search space of neural architectures while also requiring significant manual effort from domain experts. In this work, we develop techniques that enable efficient NAS in a significantly larger design space. To accomplish this, we propose to perform NAS in an abstract search space of program properties. Our key insights are as follows: (1) the abstract search space is significantly smaller than the original search space, and (2) architectures with similar program properties also have similar performance; thus, we can search more efficiently in the abstract search space. To enable this approach, we also propose a novel efficient synthesis procedure, which accepts a set of promising program properties, and returns a satisfying neural architecture. We implement our approach, $\alpha$NAS, within an evolutionary framework, where the mutations are guided by the program properties. Starting with a ResNet-34 model, $\alpha$NAS produces a model with slightly improved accuracy on CIFAR-10 but 96% fewer parameters. On ImageNet, $\alpha$NAS is able to improve over Vision Transformer (30% fewer FLOPS and parameters), ResNet-50 (23% fewer FLOPS, 14% fewer parameters), and EfficientNet (7% fewer FLOPS and parameters) without any degradation in accuracy.