Title :
Towards close-to-nature neural networks
Author :
Stoop, R. ; Bunimovich, L.A.
Author_Institution :
Inst. fur Neuroinf., ETHZ, Switzerland
Abstract :
When regularly spiking rat cortical cells perturbed by periodic inhibition, for a set of positive measures of specific ratios between stimulation and self-oscillation frequency, the resulting spiking pattern is chaotic. Contrary to earlier speculations, these connections do not desynchronize the network. The optimal network performance is characterized by a transition from local chaos to global chaos dominance. When a phase-coincidence detection algorithm is applied, quick convergence towards nontrivial phase patterns is observed. Distinct “sensory” inputs to the network are reflected in localized, input-specific differences of the observed attractors
Keywords :
brain models; cellular biophysics; chaos; neural nets; neurophysiology; nonlinear dynamical systems; attractors; chaotic response; close-to-nature neural networks; global chaos; local chaos; localized input-specific differences; nonlinear dynamics; nontrivial phase patterns; optimal network performance; periodic inhibition; periodic inhibitory stimulation; phase-coincidence detection algorithm; quick convergence; rat cortical cells; self-oscillation frequency; sensory inputs; spiking pattern; stimulation; transition; Artificial neural networks; Biological neural networks; Biological system modeling; Brain modeling; Chaos; Displays; Frequency; Glass; In vitro; Neural networks;
Conference_Titel :
Electronics, Circuits and Systems, 1999. Proceedings of ICECS '99. The 6th IEEE International Conference on
Conference_Location :
Pafos
Print_ISBN :
0-7803-5682-9
DOI :
10.1109/ICECS.1999.812351
