Regensburg 2007 – wissenschaftliches Programm

Bereiche | Tage | Auswahl | Suche | Downloads | Hilfe

DY: Fachverband Dynamik und Statistische Physik

DY 2: Time-delayed feedback and neural networks

DY 2.4: Vortrag

Montag, 26. März 2007, 11:15–11:30, H3

Oscillatory associative memory in an electrochemical system — •Robert Hölzel and Katharina Krischer — TU München, Physik Department (E 19), James Franck Str. 85748 Garching

A system of globally coupled Kuramoto oscillators with a weak time-dependent coupling can act as a Hopfield-like neural network with the pattern information stored in the relative phase shifts of the oscillators [Hoppensteadt FC Izhikevich EM, PHYSICAL REVIEW LETTERS 82 (14): 2983-2986 APR 5 1999 ].

We investigate the Hopfield-like properties of a realistic network of electrochemical oscillators with a time-dependent, weak global coupling through the electric field. Averaged evolution equations for the phase shifts are derived based on phase response curves.

Our results indicate that pattern recognition with two memorized patterns is possible in the realistic system without qualitative modification of constraints compared to the Kuramoto model. However, further refinement of the time-dependent coupling is necessary for general pattern recognition with more than two memorized patterns.