BriefGPT.xyz
Nov, 2015
没有卷积能走多远：改进全连接网络
How far can we go without convolution: Improving fully-connected networks
HTML
PDF
Zhouhan Lin, Roland Memisevic, Kishore Konda
TL;DR
探讨改进全连接网络的性能的方法，提出了线性瓶颈层和非监督预训练方法的影响，探索它们如何通过提高梯度流和减少稀疏性来改进网络，展示全连接网络在置换不变的CIFAR-10任务上达到了约70%的分类准确率，添加变形数据后达到了78%的准确率，接近卷积网络的水平。
Abstract
We propose ways to improve the performance of
fully connected networks
. We found that two approaches in particular have a strong effect on performance:
linear bottleneck layers
and
→