BriefGPT.xyz
Nov, 2017
深度重连：训练极度稀疏的深度网络
Deep Rewiring: Training very sparse deep networks
HTML
PDF
Guillaume Bellec, David Kappel, Wolfgang Maass, Robert Legenstein
TL;DR
DEEP R是一种基于后验随机抽样的算法，可以直接训练稀疏连接的神经网络，可以在标准基准测试任务中训练非常稀疏的前馈和递归神经网络，性能损失很小。
Abstract
neuromorphic hardware
tends to pose limits on the connectivity of deep networks that one can run on them. But also generic hardware and software implementations of
deep learning
run more efficiently on
→