January 2018
Beginner to intermediate
284 pages
8h 35m
English
The output gate decides what we’re going to output, that is, selectively decide what part of the current cell state we want to output as the new hidden state/output/prediction. Again, the sigmoid node is used to generate the filtering vector from
(decides what parts of the current cell state would be selected). Then, the current cell state
is put through a tanh (for the purpose of squashing the value to between -1 and 1), and multiplied by the output of the sigmoid gate. Then we get our final output .
Read now
Unlock full access