Pulse-type hardware chaotic neuron model and its bifurcation phenomena
Kazutaka Someya, Hidekazu Shinozaki, Yoshifumi Sekine · Electronics and Communications in Japan (Part II Electronics) · 1998
A number of studies have recently been made on hardware realization of a biological neuron for application of the information processing functions of biological neural networks to engineering problems. Especially, it is suggested that new information processing structures can be realized in artificial neural networks with the chaotic neuron model. But a pulse-type hardware chaotic neuron model realized by electronic circuits has not been proposed. In our previous studies, we have proposed the Λ type neuron model with a negative resistance circuit composed of two complementary FETs. The Λ type neuron model can be consumed by a simple circuit with a design geared to easy IC fabrication. Also, we have reported that the Λ type neuron model exhibited continuous firing for a constant stimulus current and so on. In this paper, we show that the Λ type neuron model with a synaptic section has all features of the chaotic neuron model and that the pulse-type hardware chaotic neuron model can be realized. Furthermore, we show that the proposed model has three chaotic regions in the bifurcation diagram for periodic pulse train stimulation, and we make clear the bifurcation route and the return map in each chaotic region. © 1998 Scripta Technica, Electron Comm Jpn Pt 2, 81(2): 55–62, 1998