it is well-known that in a feedback network the output signals from a layer can be given as input to neurons in its own layer or to layers before it. But, is it right to consider feedback to the input layer neurons also?
Liangliang Shi
Surely you can do this since each layer of the input is the output of another layer.
0 votes
0 thanks
Issa Ayoub
If I understood your question thoroughly, you would like to feedback the output of an RNN (if that's after n layers), into the first layer? If yes, then this is equivalent to not having just a fully connected layer between the inputs and the hidden activation units, instead, you have a whole network, of n layers between the model's inputs and the RNN hidden activations. Hope this helps!!
0 votes
0 thanks
- Badges
- Science topic
More B.K. Tripathy's questions See All
Similar questions and discussions