1. Learning long-term dependencies with gradient descent is difficult
- Author
-
Paolo Frasconi, Patrice Y. Simard, and Yoshua Bengio
- Subjects
Vanishing gradient problem ,Dynamical systems theory ,Artificial neural network ,Computer Networks and Communications ,Computer science ,business.industry ,Numerical analysis ,Recurrent neural nets ,General Medicine ,Machine learning ,computer.software_genre ,Computer Science Applications ,Term (time) ,Recurrent neural network ,Artificial Intelligence ,Face (geometry) ,Pattern recognition (psychology) ,Artificial intelligence ,business ,Gradient descent ,Gradient method ,computer ,Software - Abstract
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. >
- Published
- 1994
- Full Text
- View/download PDF