Abstract
Two compact analog CMOS synaptic circuits with in situ Hebbian learning have been developed and used to construct a Mean Field Network (MFN) which is a deterministic version of a Boltzmann machine. This network, consisting of 25 neurons and 625 synapses with local learning and weight storage is currently being fabricated in 3μm CMOS. Our investigations show that neural network architectures,such as the MFN can be constructed from highly non-ideal analog CMOS components because the adaptive ability of neural net architectures with learning allows the network to compensate for device variations.
This is a preview of subscription content, log in via an institution.
Buying options
Tax calculation will be finalised at checkout
Purchases are for personal use only
Learn about institutional subscriptionsPreview
Unable to display preview. Download preview PDF.
References
Card, H.C. and Moore, W.R., “VLSI devices and circuits for neural networks”, Int. J. Neural Systems, 1989, Vol. 1, pp. 149–165.
Clark, J.T., “An analog CMOS implementation of a self organizing feed-forward network”, Proc. Int. Joint Conf. on Neural Networks, IJCNN-WASH 90, M. Caudill, editor, 1990, Vol. 1, pp. 118–121.
Hinton, G.E., “Deterministic Boltzmann learning performs steepest descent in weight space”, Neural Computation, Vol. 1, No. 1, 1989, pp. 143–150.
Kohonen, T., “An introduction to neural computing”, Neural Networks, 1988, 1, pp. 3–16.
Mead, C.A., Analog VLSI and Neural Systems, 1988, Reading: Addison-Wesley
Murray, A.F., “Silicon implementation of neural networks”, Proc. First IEE Int. Conf. onArtificial Neural Networks, 1989, pp. 27–32.
Peterson, C. and Hartman, A., “Explorations of the Mean Field Theory Learning Algorithm”, Neural Networks, Vol. 2, pp. 475–494, 1989.
Schwartz, D.B., Howard, R.E., and Hubbard, W.E., “A programmable analog neural network chip”, IEEE J. Solid St. Ccts., 1989, 24, pp. 313–319
Shimabukuro, R.L., Reedy, R.E., Garcia, G.A., “Dual-polarity nonvolatile MOS analog memory (MAM) cell for neural-type circuitry”, Electronic Letters, 1988, 24, pp. 1231–1232.
Author information
Authors and Affiliations
Editor information
Editors and Affiliations
Rights and permissions
Copyright information
© 1991 Springer Science+Business Media New York
About this chapter
Cite this chapter
Schneider, C., Card, H. (1991). Analog VLSI Models of Mean Field Networks. In: Delgado-Frias, J.G., Moore, W.R. (eds) VLSI for Artificial Intelligence and Neural Networks. Springer, Boston, MA. https://doi.org/10.1007/978-1-4615-3752-6_18
Download citation
DOI: https://doi.org/10.1007/978-1-4615-3752-6_18
Publisher Name: Springer, Boston, MA
Print ISBN: 978-1-4613-6671-3
Online ISBN: 978-1-4615-3752-6
eBook Packages: Springer Book Archive