BriefGPT.xyz
Jun, 2016
关于循环神经网络的乘积积分
On Multiplicative Integration with Recurrent Neural Networks
HTML
PDF
Yuhuai Wu, Saizheng Zhang, Ying Zhang, Yoshua Bengio, Ruslan Salakhutdinov
TL;DR
本文提出了一种称为乘积积分（MI）的结构设计来改进递归神经网络（RNN），并在几个任务中对不同RNN模型的学习行为进行了实验评估。实验结果表明，MI可以显著提高现有的许多RNN模型的性能。
Abstract
We introduce a general and simple structural design called
multiplicative integration
(MI) to improve
recurrent neural networks
(RNNs). MI changes the way in which information from difference sources flows and is
→