Title | ||
---|---|---|
FBNetV2: Differentiable Neural Architecture Search for Spatial and Channel Dimensions |
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. |
Year | DOI | Venue |
---|---|---|
2020 | 10.1109/CVPR42600.2020.01298 | CVPR |
DocType | Citations | PageRank |
Conference | 8 | 0.47 |
References | Authors | |
24 | 12 |
Name | Order | Citations | PageRank |
---|---|---|---|
Alvin Wan | 1 | 15 | 2.59 |
Xiaoliang Dai | 2 | 126 | 8.58 |
Peizhao Zhang | 3 | 252 | 12.47 |
Zijian He | 4 | 11 | 1.85 |
Yuandong Tian | 5 | 703 | 43.06 |
Saining Xie | 6 | 231 | 12.45 |
Bichen Wu | 7 | 96 | 7.15 |
Matt C. Yu | 8 | 89 | 8.62 |
Tao Xu | 9 | 187 | 11.21 |
Chen Kan | 10 | 131 | 9.28 |
Peter Vajda | 11 | 80 | 5.14 |
Joseph E. Gonzalez | 12 | 2219 | 102.68 |