Skip to content

Latest commit

 

History

History
15 lines (7 loc) · 959 Bytes

PathNet:Evolution-Channels-Gradient-Descent-in-Super-Neural-Networks.md

File metadata and controls

15 lines (7 loc) · 959 Bytes

PathNet

paper: PathNet: Evolution Channels Gradient Descent in Super Neural Networks

===

DeepMind论文展示通用AI是什么样子

Deepmind连续学习(序列学习)三连弹

三篇论文分别是:Progressive Neural Networks(https://arxiv.org/pdf/1606.04671.pdf), PathNet: Evolution Channels Gradient Descent in Super Neural Networks(https://arxiv.org/pdf/1701.08734.pdf) 和 Overcoming catastrophic forgetting in neural networks(后面我就简写EWC了)(http://www.pnas.org/content/early/2017/03/13/1611835114.full.pdf).