A traditional approach to initialization in deep neural networks (DNNs) is to sample the network weights randomly for preserving the variance of pre-activations. On the other hand, several studies show that during the training process, the distribution of stochastic gradients can be heavy-tailed especially for small batch sizes. In this case, weights and therefore pre-activations can be modeled with a heavy-tailed distribution that has an infinite variance but has a finite (non-integer) fractional moment of order $s$ with $s<2$. Motivated by this fact, we develop initialization schemes for fully connected feed-forward networks that can provably preserve any given moment of order $s \in (0, 2]$ over the layers for a class of activations including ReLU, Leaky ReLU, Randomized Leaky ReLU, and linear activations. These generalized schemes recover traditional initialization schemes in the limit $s \to 2$ and serve as part of a principled theory for initialization. For all these schemes, we show that the network output admits a finite almost sure limit as the number of layers grows, and the limit is heavy-tailed in some settings. This sheds further light into the origins of heavy tail during signal propagation in DNNs. We prove that the logarithm of the norm of the network outputs, if properly scaled, will converge to a Gaussian distribution with an explicit mean and variance we can compute depending on the activation used, the value of s chosen and the network width. We also prove that our initialization scheme avoids small network output values more frequently compared to traditional approaches. Furthermore, the proposed initialization strategy does not have an extra cost during the training procedure. We show through numerical experiments that our initialization can improve the training and test performance.


翻译:在深神经网络(DNNS)中,一种传统的初始化方法是随机抽样网络权重,以保存启动前的差别。另一方面,一些研究显示,在培训过程中,特别小批量尺寸,可大量压缩沙沙梯度的分布。在此情况下,权重和因此的启动前,可模拟出一个大细细细的分布模式,该分布模式具有无限差异,但具有一定(非整数)的(非整数)直线化时间点,其价值为 <2美元。受此事实的驱动,我们为完全连接的进化前网络开发初始化计划,该网络的初始化计划可以保存任何特定时刻的 $ / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / /

1
下载
关闭预览

相关内容

Networking:IFIP International Conferences on Networking。 Explanation:国际网络会议。 Publisher:IFIP。 SIT: http://dblp.uni-trier.de/db/conf/networking/index.html
神经网络的拓扑结构,TOPOLOGY OF DEEP NEURAL NETWORKS
专知会员服务
32+阅读 · 2020年4月15日
机器学习入门的经验与建议
专知会员服务
92+阅读 · 2019年10月10日
已删除
将门创投
5+阅读 · 2019年4月4日
A Technical Overview of AI & ML in 2018 & Trends for 2019
待字闺中
17+阅读 · 2018年12月24日
强化学习 cartpole_a3c
CreateAMind
9+阅读 · 2017年7月21日
Arxiv
6+阅读 · 2018年10月3日
Arxiv
3+阅读 · 2018年8月17日
VIP会员
相关资讯
已删除
将门创投
5+阅读 · 2019年4月4日
A Technical Overview of AI & ML in 2018 & Trends for 2019
待字闺中
17+阅读 · 2018年12月24日
强化学习 cartpole_a3c
CreateAMind
9+阅读 · 2017年7月21日
Top
微信扫码咨询专知VIP会员