Abstract
Control of spatiotemporal chaos in a neural network with discrete time and continuous state variables is investigated. The chaos control is performed with the knowledge of only a part of the target information in the memory patterns. The success rate for the pattern associations and the dependence of the search time on the sampling number in the proposed chaos neural network are studied. By the introduction of the reinforcement factor in the learning process, the recognition rate of the network can be much enhanced. Random and regular samplings of the pattern for the control are tested and the successful results of the associations are demonstrated. The chaotic behavior and recalling ability of the system are evaluated based on the analysis of the Lyapunov spectrum of the network.
- Received 18 December 1995
DOI:https://doi.org/10.1103/PhysRevE.53.4502
©1996 American Physical Society