BriefGPT.xyz
May, 2017
循环加性网络
Recurrent Additive Networks
HTML
PDF
Kenton Lee, Omer Levy, Luke Zettlemoyer
TL;DR
本文提出了一种基于输入和上一状态的门限组件加法构成的递归神经网络RANs,与LSTM在基准语言建模问题上表现相当,表明LSTM中的非线性计算在此类问题上并非必需品,而门限完成了更多的计算工作。
Abstract
We introduce
recurrent additive networks
(RANs), a new
gated rnn
which is distinguished by the use of purely additive latent state updates. At every time step, the new state is computed as a gated component-wise
→