Learning Long-Term Dependencies with Gradient
Descent is Difficult
Yoshua Bengioy, Patrice Simardy, and Paolo Frasconiz
yAT&T Bell Laboratories
zDip. di Sistemi e Informatica, Universit?a di Firenze
Recurrent neural networks can be used to map input sequences to output sequences, such as for recognition, production or prediction problems. However, practical difficulties have been reported in training recurrent neural networks to perform tasks in which the temporal contingencies present in the input/output sequences span long intervals. We show why gradient based learning algorithms face an increasingly difficult problem as the duration of the dependencies to be captured increases. These results expose a trade-off between efficient learning by gradient descent and latching on information for long periods. Based on an understanding of this problem, alternatives to standard gradient descent are considered.
Paper to appear in the special issue on Recurrent Networks of the IEEE Transactions on Neural Networks