A Learning Algorithm for Continually Running Fully Recurrent Neural Networks
- 1 June 1989
- journal article
- research article
- Published by MIT Press in Neural Computation
- Vol. 1 (2) , 270-280
- https://doi.org/10.1162/neco.1989.1.2.270
Abstract
The exact form of a gradient-following learning algorithm for completely recurrent networks running in continually sampled time is derived and used as the basis for practical algorithms for temporal supervised learning tasks. These algorithms have (1) the advantage that they do not require a precisely defined training interval, operating while the network runs; and (2) the disadvantage that they require nonlocal communication in the network being trained and are computationally expensive. These algorithms allow networks having recurrent connections to learn complex tasks that require the retention of information over time periods having either fixed or indefinite length.This publication has 5 references indexed in Scilit:
- Learning State Space Trajectories in Recurrent Neural NetworksNeural Computation, 1989
- Dynamics and architecture for neural computationJournal of Complexity, 1988
- A self-optimizing, nonsymmetrical neural net for content addressable memory and pattern recognitionPhysica D: Nonlinear Phenomena, 1986
- Neural networks and physical systems with emergent collective computational abilities.Proceedings of the National Academy of Sciences, 1982
- Optimization of time-varying systemsIEEE Transactions on Automatic Control, 1965