In recent years,the neural network is used in many fields. Furthermore , neural networks with a chaotic phenomenon is also begin studied. While McCulloch and Pitts introduced the neural network in 1943 , the study and research of the Chaotic Neural Network has only now begun to prosper.
In our laboratory, the learning methods of a network have been studied based upon the Hebbian rule. In this study, we experimented with the On-Demand Learning Method applied by the Hebbian rule. The value of the connection weight which is a learning parameter was made into the variable. It investigated how efficiency changes with respect to connection weight. In addition an experiment which resets an internal state was also conducted. The function to reset the inside is important for the learning of a network. It would be difficult to learn if connection weight was not a suitable value as a result of an experiment. Until now , the value of from our laboratory used by the connection weight has been 0.05. This value was suitable.
競合