Deep Learning without Backpropagation – Tutorial: DeepMind’s Synthetic Gradients

Deepmind提出Synthetic Gradient來取代Backpropagation,藉由獨立優化各層的Synthetic Gradient Generator能提高訓練的速度。此篇Tutotrial帶領你快速理解原理,還有簡單的實作程式碼可以玩唷!

[Link] https://iamtrask.github.io/2017/03/21/synthetic-gradients/ (Tutorial)

[Paper] https://arxiv.org/pdf/1608.05343.pdf (Decoupled Neural Interfaces using Synthetic Gradients)

[Paper] https://arxiv.org/abs/1703.00522 (Understanding Synthetic Gradients and Decoupled Neural Interfaces)

Share the joy
  •  
  •  
  •  
  •  
  •  
  •  
  •  
  •  
  •  
  •  

近期文章

近期迴響

彙整

分類

其它

leoyang Written by:

Be First to Comment

發表迴響

你的電子郵件位址並不會被公開。 必要欄位標記為 *