Complicated and computational dynamics of spatio-temporal neurochaos

Deterministic chaos is studied in relation to chaotic neurodynamics and its possible applications to artificial neurocomputations. First, chaotic dynamics in real nerve membranes is demonstrated not only numerically with nerve equations such as the Hodgkin-Huxley equations and the FitzHugh-Nagumo equations but also experimentally with squid giant axons. The neurochaos in the level of single neurons is qualitatively described with a simple neuron model of one-dimensional mapping derived on the basis of physiological properties of the nerve membranes. Second, complicated spatio-temporal neurochaos in neural networks composed of such chaotic neuron models is analysed from the viewpoint of artificial neurocomputations such as associative dynamics and optimization dynamics. Last, the model of the chaotic neural networks is further extended to transient chaotic neural networks with chaotic simulated annealing and asynchronous chaotic neural networks with temporal coding by coincidence detection and continuous interspike intervals.