International Symposium on Nonlinear Theory and its Applications


Session Number:B1L-E



A mathematical-structure-based aVLSI silicon neuron model

Takashi Kohno,  Kazuyuki Aihara,  


Publication Date:2010/9/5

Online ISSN:2188-5079


PDF download (1017.5KB)

The mathematical structures under the conductance-based neuron models have been studied extensively from the perspective of the nonlinear dynamics and bifurcation theory. We proposed to design silicon neuron models by re-constructing the topological structures in the phase portraits and the bifurcation diagrams of the conductance-based neuron models utilizing device-native curves. It not only allows us to design simple circuitry retaining the neuronal dynamics but also provides effective procedures to determine the parameter voltages applied to operate the circuits. An analog Very-Large-Scale Integration (aVLSI) silicon neuron model that mimics the math- ematical structures in two groups of bursting neurons was designed based on this idea. The results of the theoretical model and HSpice simulations are reported.