Accession Number:

AD0415223

Title:

DYNAMIC INFORMATION HANDLING IN NEURISTOR SYSTEMS,

Descriptive Note:

Corporate Author:

STANFORD RESEARCH INST MENLO PARK CA COMPUTER TECHNIQUES LAB

Personal Author(s):

Report Date:

1963-01-01

Pagination or Media Count:

43.0

Abstract:

Neuristor is the name given to any of a class of devices that exhibit attenuationless propagation with refractoriness, and includes the familiar nerve axon as a special case. Although initially proposed as a solution to the potential problems of attenuation in the interconnections of highly miniaturized electronic systems, the information processing capabilities of neuristor networks are diverse and interesting. Indeed it has been shown that arbitrarily complex logic activity can be achieved with systems composed exclusively of the simplest kind of lines and of two elemental modes of connection between them. These basic capabilities are briefly reviewed and various extensions to these capabilities are considered. Several specific examples of uniquely neuristor configurations are treated these include, for example, the design of networks using probabilis tic gates, delay-separable networks, and the use of non-binary signal representations. In par ticular, the study of pulse-train interactions leads to techniques for processing analog infor mation, as well as digital information, which are suggestive of methods by which similar processing might be accomplished in neural systems. Author

Subject Categories:

Distribution Statement:

APPROVED FOR PUBLIC RELEASE