A Study of Back Propagation Learning with PeriodicChaos Neurons

In this paper, we shall propose an error back propagation scheme with a periodicchaos neuron model. The present learning model is properly characterized in terms of periodicchaos neurons to involve chaotic dynamics as well as external or autonomous control of theperiodicity. In practice, one may confirm that there exists a close relationship between the learningprocess and chaotic dynamics, and that chaotic dynamics promote the learning speed with acertain success rate, as seen in the conventional nonchaotic models.