Inspired by the harmony between the basic functional elements of biological neural networks and their natural operating media, we have been seeking for ways to implement artificial neural networks (ANNs) using the intrinsic functionality of the most commonly available devices in an electronics technology, in contrast to the method of hardware-compilation of software-simulation modules. In the case of MOS technology, we employ a quadratic functional equation similar to that found in standard MOS transistors to implement synapses in ANNs. A structure has been proposed in to implement a MOS device with externally-controllable threshold voltage to be employed as a synapse. In the present work, we develop and compare practical architectures within which these synapses can be utilized optimally. A simulator and proper training algorithms have been developed to simulate different hardware-based architectures.<<ETX>>
[1]
Dennis J. Volper,et al.
Quadratic function nodes: Use, structure and training
,
1990,
Neural Networks.
[2]
Tadashi Shibata,et al.
A functional MOS transistor featuring gate-level weighted sum and threshold operations
,
1992
.
[3]
John Moody,et al.
Fast Learning in Networks of Locally-Tuned Processing Units
,
1989,
Neural Computation.
[4]
Carver Mead,et al.
Analog VLSI and neural systems
,
1989
.
[5]
Geoffrey E. Hinton,et al.
Learning internal representations by error propagation
,
1986
.
[6]
R. Palmer,et al.
Introduction to the theory of neural computation
,
1994,
The advanced book program.