Improve Ranking Correlation of Super-net through Training Scheme from One-shot NAS to Few-shot NAS
This work addresses a specific bottleneck in NAS for researchers and practitioners, offering an incremental improvement to enhance ranking correlation in lightweight architecture search.
The paper tackles the problem of poor ranking consistency among subnets in one-shot neural architecture search (NAS) due to weight-sharing interference, by proposing a step-by-step training scheme that transitions from one-shot to few-shot NAS. The result is a method that achieved 4th place in the CVPR2022 Lightweight NAS Challenge Track1.
The algorithms of one-shot neural architecture search(NAS) have been widely used to reduce computation consumption. However, because of the interference among the subnets in which weights are shared, the subnets inherited from these super-net trained by those algorithms have poor consistency in precision ranking. To address this problem, we propose a step-by-step training super-net scheme from one-shot NAS to few-shot NAS. In the training scheme, we firstly train super-net in a one-shot way, and then we disentangle the weights of super-net by splitting them into multi-subnets and training them gradually. Finally, our method ranks 4th place in the CVPR2022 3rd Lightweight NAS Challenge Track1. Our code is available at https://github.com/liujiawei2333/CVPR2022-NAS-competition-Track-1-4th-solution.