FBNetV2: Differentiable Neural Architecture Search for Spatial and Channel Dimensions

TL;DR

This paper introduces DMaskingNAS, a memory-efficient differentiable neural architecture search method that significantly expands the search space to include spatial and channel dimensions, leading to state-of-the-art models with less search cost.

Contribution

The paper proposes DMaskingNAS, a novel DNAS variant supporting large search spaces over spatial and channel dimensions with constant memory, enabling efficient search for high-performance neural networks.

Findings

DMaskingNAS expands search space by up to 10^14 times.

FBNetV2 models achieve state-of-the-art accuracy with fewer FLOPs.

Models are found with up to 421x less search cost.

Abstract

Differentiable Neural Architecture Search (DNAS) has demonstrated great success in designing state-of-the-art, efficient neural networks. However, DARTS-based DNAS's search space is small when compared to other search methods', since all candidate network layers must be explicitly instantiated in memory. To address this bottleneck, we propose a memory and computationally efficient DNAS variant: DMaskingNAS. This algorithm expands the search space by up to $10^{14}\times$ over conventional DNAS, supporting searches over spatial and channel dimensions that are otherwise prohibitively expensive: input resolution and number of filters. We propose a masking mechanism for feature map reuse, so that memory and computational costs stay nearly constant as the search space expands. Furthermore, we employ effective shape propagation to maximize per-FLOP or per-parameter accuracy. The searched FBNetV2s yield state-of-the-art performance when compared with all previous architectures. With up to 421$\times$ less search cost, DMaskingNAS finds models with 0.9% higher accuracy, 15% fewer FLOPs than MobileNetV3-Small; and with similar accuracy but 20% fewer FLOPs than Efficient-B0. Furthermore, our FBNetV2 outperforms MobileNetV3 by 2.6% in accuracy, with equivalent model size. FBNetV2 models are open-sourced at https://github.com/facebookresearch/mobile-vision.