【发布时间】:2017-04-06 20:57:59
【问题描述】:
感谢您阅读这篇文章!
这里是给 RNN 爱好者的快速问题:
我知道在反向传播中(BPPT),至少有3个步骤:
For each element in a sequence :
Step 1 - Compute 'error ratio' of each neuron, from upper layer to lower layer.
Step 2 - Compute a 'weight delta' for each weight (X) using the error ratio mentionned in step 1, and push it into an array
After sequence is finished :
Step 3 - Sum all weight deltas of weight (X) and add it to current value of weight (X)
我现在正在尝试从此处找到的文档中实现发条 RNN (CW RNN): http://jmlr.org/proceedings/papers/v32/koutnik14.pdf
据我了解,隐藏层中的每个“模块”都有相同数量的神经元,只是一个不同的时钟。
CW RNN 的前向传递看起来非常简单直观。
然而,对于向后传球,那就另当别论了。
引用文档:
The backward pass of the error propagation is similar to
SRN as well. The only difference is that the error propagates
only from modules that were executed at time step t. The
error of non-activated modules gets copied back in time
(similarly to copying the activations of nodes not activated
at the time step t during the corresponding forward pass),
where it is added to the back-propagated error.
这就是我困惑的地方。
上述哪些反向传播步骤应用于隐藏层中的未激活模块?
(clock MOD timestep != 0 的模块)
step1 、 step2 还是两者兼有?
再次感谢您的帮助!
【问题讨论】:
-
据我了解,它是信噪比,在我看来,这是一个我还没有完全理解的概念
-
我的评论是:
what is SRN。来自论文:SRN = simple RNN(我懒得看论文)
标签: neural-network recurrent-neural-network backpropagation