Next: Unfolding in time
Up: The Architecture and the
Previous: The Architecture and the
The off-line version of the algorithm would wait for the end of
an episode to compute the final change of as the sum of all
changes computed at each time step. The on-line version
changes at every time step, assuming that is small
enough to avoid instabilities [Williams and Zipser, 1989].
An interesting property of the on-line version is that we
do not have to specify episode boundaries (`all episodes
blend into each other' [Williams and Zipser, 1989]).
Back to Recurrent Neural Networks page