Conference Paper Implementation of Chaos Neural Network which Generates Multi-Subseries with Different Periods

YOSHIDA, Hitoaki  ,  SASAKI, Mitsuaki  ,  MURAKAMI, Takeshi

A chaos neural network (B-6nn) which generates three independent subseries has been implemented. The sub-series afford different chaos orbits, respectively. The results of NIST SP800-22 tests also have been fine, if pseudo-random numbers are extracted from the lower-24-bit of an output in B-6nn. The whole period of outputs of B-6nn has been estimated ca. 1.58×10 22. Compared with the whole period of the conventional chaos neural network (C-4nn) which consists of 4 neurons 10 16-10 18, the whole period of B-6nn has considerably improved. The method will be applied to multi-subseries more than three subseries in future work.

Number of accesses :  

Other information