LSTM是一种递归神经网络，通过独特的门控机制解决了梯度消失问题，能有效保持长期记忆。文章介绍了LSTM的架构，并详细阐述了其内部公式及导数推导，帮助读者深入理解LSTM的工作原理。
Long Short-term memory is a kind of recurrent neural networks. It has additional gates and its' architecture is bellow:
This architecture can avoid the gradient vanishing and keep the long-term memory, so this is a very powerful neural network.
We can get these formulations:
The symbol denotes the indexs of all units,
denotes
the output of unit
.
First of all, we define the truncated item:
In the later derivation, we may use the to denote
for
conventional.
Secondly, we will derive the derivative.
扫一扫
343
被折叠的 条评论
为什么被折叠?
到【灌水乐园】发言
