Programmable dynamics in an analog VLSI neuromorph

The architecture of an analog VLSI neuromorph which can receive and process large numbers of pulsatile inputs via a spatially extensive artificial dendritic tree is described. The artificial dendritic tree is a hybrid VLSI circuit which has distributed over its length synapses of three types: hyperpolarizing, depolarizing, and shunting. Because of the distributed nature of the circuit and its natural dynamics, input signals are segregated both spatially and temporally. In addition, neuromorph dynamics is programmable over a wide range which permits one to match neuromorph dynamical response with application dynamics. Experimental results are presented that show "membrane" response of a directionally selective neuromorph preserved over four orders of magnitude in target speed as neuromorph dynamics is correspondingly changed.<<ETX>>

[1]  J. G. Elias Silicon Dendritic Trees , 1994 .

[2]  T. Poggio,et al.  Biophysics of Computation: Neurons, Synapses and Membranes , 1984 .

[3]  T. Poggio,et al.  Nonlinear interactions in a dendritic tree: localization, timing, and role in information processing. , 1983, Proceedings of the National Academy of Sciences of the United States of America.

[4]  John G. Elias,et al.  Artificial Dendritic Trees , 1993, Neural Computation.

[5]  T. Poggio,et al.  A synaptic mechanism possibly underlying directional selectivity to motion , 1978, Proceedings of the Royal Society of London. Series B. Biological Sciences.