[3I4-GS-7a] 画像音声メディア処理：基礎

[3I4-GS-7a-01] Differentiable Architecture Search Using Stable Rank of Convolutional Layers

〇Kengo Machida¹, Kuniaki Uto¹, Koichi Shinoda¹, Taiji Suzuki^2,3 (1. Tokyo Institute of Technology, 2. Graduate School of Information Science and Technology, The University of Tokyo, Japan, 3. Center for Advanced Intelligence Project, RIKEN, Japan)

Keywords:AI, Deep Learning, Neural Architecture Search, DARTS; Differentiable Architecture Search, Stable Rank

We propose Neural Architecture Search method MSR-DARTS, which uses the stable rank of a convolutional layer for selecting the optimal network structure.
We evaluate MSR-DARTS on CIFAR10 and ImageNet dataset.
It achieves an error rate of 2.80% on CIFAR10 and a top-1 error rate of 24.0% of ImageNet.