Inductive Inference Scheme Based on Spontaneous Fluctuation of a Probabilistic Neural Network
Masahiro Agu Yamanaka · Japanese Journal of Applied Physics · 1993
It is shown that spontaneous fluctuation of the neural excitation pattern of a probabilistic neural network can be the source or seed for generation of information in order to perform a kind of inductive inference. A primary model of the inductive inference scheme is proposed and applied to a pattern restoration problem. Results of computer simulation suggest the validity of the proposed model.