BriefGPT.xyz
Mar, 2020
如何训练超网络：基于权重共享的NAS训练启发式分析
How to Train Your Super-Net: An Analysis of Training Heuristics in Weight-Sharing NAS
HTML
PDF
Kaicheng Yu, Rene Ranftl, Mathieu Salzmann
TL;DR
通过系统评估常用的超网络训练启发式方法和超参数，发现其中一些启发式方法会对超网络和独立性能之间的相关性产生负面影响，并证明了某些超参数和体系结构选择的强烈影响。
Abstract
weight sharing
promises to make
neural architecture search
(NAS) tractable even on commodity hardware. Existing methods in this space rely on a diverse set of
→