Title :
Current-mode subthreshold MOS implementation of the Herault-Jutten autoadaptive network
Author :
Cohen, Marc H. ; Andreou, Andreas G.
Author_Institution :
Nat. Inst. of Health, Bethesda, MD, USA
fDate :
5/1/1992 12:00:00 AM
Abstract :
The authors explore translinear circuits in subthreshold MOS technology and current-mode design techniques for the implementation of neuromorphic analog network processing. The architecture, also known as the Herault-Jutten network, performs an independent component analysis and is essentially a continuous-time recursive linear adaptive filter. Analog I/O interface, weight coefficients, and adaptation blocks are all integrated on the chip. A small network with six neurons and 30 synapses was fabricated in a 2-μm double-polysilicon, double-metal n-well CMOS process. Circuit designs at the transistor level yield area-efficient implementations for neurons, synapses, and the adaptation blocks. The authors discuss the design methodology and constraints as well as test results from the fabricated chips
Keywords :
CMOS integrated circuits; VLSI; linear integrated circuits; neural nets; 2 micron; Herault-Jutten network; adaptation blocks; analogue I/O interface; area-efficient implementations; autoadaptive network; continuous-time; current-mode design techniques; design methodology; double-metal; double-polysilicon; independent component analysis; n-well CMOS process; neuromorphic analog network processing; neurons; recursive linear adaptive filter; subthreshold MOS implementation; synapses; translinear circuits; weight coefficients; Adaptive filters; CMOS process; Circuit synthesis; Circuit testing; Design methodology; Independent component analysis; Integrated circuit technology; Integrated circuit yield; Neuromorphics; Neurons;
Journal_Title :
Solid-State Circuits, IEEE Journal of