Back to the Previous subjectBackpropagation Through Time |
|---|
|
While Recurrent Backpropagation tries to minimize the network's error after presenting each training input/output pair, backpropagation through time will look at the performance of the network as an approximation of a certain function. For this reason, it looks at the mapping of all input/output pairs before making a decision about how to change the weights.
|
Back to the Table of Content
To the Next Subject