SCARLET-NAS: Bridging the Gap between Stability and Scalability in Weight-sharing Neural Architecture Search
arXiv:1908.06022
Abstract
To discover powerful yet compact models is an important goal of neural architecture search. Previous two-stage one-shot approaches are limited by search space with a fixed depth. It seems handy to include an additional skip connection in the search space to make depths variable. However, it creates a large range of perturbation during supernet training and it has difficulty giving a confident ranking for subnetworks. In this paper, we discover that skip connections bring about significant feature inconsistency compared with other operations, which potentially degrades the supernet performance. Based on this observation, we tackle the problem by imposing an equivariant learnable stabilizer to homogenize such disparities. Experiments show that our proposed stabilizer helps to improve the supernet's convergence as well as ranking performance. With an evolutionary search backend that incorporates the stabilized supernet as an evaluator, we derive a family of state-of-the-art architectures, the SCARLET series of several depths, especially SCARLET-A obtains 76.9% top-1 accuracy on ImageNet. Code is available at https://github.com/xiaomi-automl/ScarletNAS.
ICCV Workshop 2021
References in corpus (6)
- MixConv: Mixed Depthwise Convolutional Kernels
- Understanding and Robustifying Differentiable Architecture Search
- NAS-Bench-101: Towards Reproducible Neural Architecture Search
- Single Path One-Shot Neural Architecture Search with Uniform Sampling
- Progressive Differentiable Architecture Search: Bridging the Depth Gap between Search and Evaluation
- FairNAS: Rethinking Evaluation Fairness of Weight Sharing Neural Architecture Search