Digitally programmable analog building blocks for the implementation of artificial neural networks

This paper describes the design, experimental characterization and behavior modeling of a homogeneous set of building blocks necessary to construct in analog hardware feed-forward artificial neural networks. A novel synapse architecture is proposed using a quasi-passive D/A (digital-to-analog) converter followed by a four-quadrant analog-digital multiplier, its main advantages are 1) increased signal input range; 2) improved area/weight resolution ratio; 3) on-chip refreshing of the weight value; and 4) serial loading the weight bits. The neurons are built using MOS (metal-oxide semiconductor) transistors operating in the saturation region and exploiting the inherent quadratic characteristics. Experimental results obtained from a demonstration prototype chip realized in a 1.2 mum double-poly, double-metal CMOS (complimentary MOS) technology show good agreement with the design specifications. A simple application of the proposed building blocks is illustrated based on the mixed-signal simulation of the corresponding behavior models constructed from the experimental characterization data.

[1]  Phillip E Allen,et al.  CMOS Analog Circuit Design , 1987 .

[2]  K. Bult,et al.  A class of analog CMOS circuits based on the square-law characteristic of an MOS transistor in saturation , 1987 .

[3]  Sigeru Omatu,et al.  A neural network based control scheme with an adaptive neural model reference structure , 1991, [Proceedings] 1991 IEEE International Joint Conference on Neural Networks.

[4]  S. Tam,et al.  An electrically trainable artificial neural network (ETANN) with 10240 'floating gate' synapses , 1990, International 1989 Joint Conference on Neural Networks.

[5]  H.P. Graf,et al.  Analog Electronic Neural Networks , 1992, ESSCIRC '92: Eighteenth European Solid-State Circuits conference.

[6]  Lawrence D. Jackel,et al.  An analog neural network processor with programmable topology , 1991 .

[7]  Carver Mead,et al.  Analog VLSI and neural systems , 1989 .

[8]  J. E. Franca,et al.  Programmable quasi-passive algorithmic digital- analogue converter , 1992 .

[9]  K. Bult,et al.  A CMOS Four-Quadrant Analog Multiplier , 1986 .

[10]  Edgar Sanchez-Sinencio,et al.  Artificial Neural Networks: Paradigms, Applications, and Hardware Implementations , 1994 .

[11]  T. Kimura,et al.  Synapse weight accuracy of analog neuro chip , 1993, Proceedings of 1993 International Conference on Neural Networks (IJCNN-93-Nagoya, Japan).

[12]  Mohammed Ismail,et al.  Analog VLSI Implementation of Neural Systems , 2011, The Kluwer International Series in Engineering and Computer Science.

[13]  S. Mitra,et al.  Handbook for Digital Signal Processing , 1993 .

[14]  F. J. Kub,et al.  Programmable analog vector-matrix multipliers , 1990 .

[15]  D. Hammerstrom,et al.  A VLSI architecture for high-performance, low-cost, on-chip learning , 1990, 1990 IJCNN International Joint Conference on Neural Networks.

[16]  Yannis Tsividis,et al.  A reconfigurable VLSI neural network , 1992 .

[17]  Bernard Widrow,et al.  Adaptive Signal Processing , 1985 .

[18]  M. Yagyu,et al.  Design, fabrication and evaluation of a 5-inch wafer scale neural network LSI composed on 576 digital neurons , 1990, 1990 IJCNN International Joint Conference on Neural Networks.