| 摘要 |
PURPOSE: To obtain a chaos recurrent neuro network by which chaos neuron is applied to a neuron of a conventional recurrent circuit network by providing chaos neuro network satisfying a specific dynamic equation. CONSTITUTION: The chaos recurrent neuro network is provided with N-sets of chaos neuro networks satisfying a dynamic equation shown in the figure. Then, each of the N-sets of the chaos neuro networks receives an external input and output signals of (N-1) sets of chaos neuro networks among the N-sets of the chaos neuro networks and conducts the operation by the dynamic equation. In the equation, Wij indicates a synapse coupling coefficient with a feedback input from a j-th neuron to an i-th neuron, Xi (t) depicts an i-th neuron output at a time (t),γi shows a time delay constant,αis a non-negative parameter and (k) is a refractory timewise attenuation constant. Zi (t) is the Xi (t) when the suffix (i) belongs to a neuron set I and indicates an external input ai (t) when the suffix (i) belongs to a set E of external inputs.
|
