Accession Number : ADA310107
Title : Neuronal Micronets as Nodal Elements.
Descriptive Note : Final rept.,
Corporate Author : YALE UNIV NEW HAVEN CT DEPT OF PSYCHOLOGY
Personal Author(s) : Brown, Thomas H.
PDF Url : ADA310107
Report Date : 16 SEP 1995
Pagination or Media Count : 6
Abstract : We have been working on developing a computationally efficient way to emulate neurons and to emulate circuits and networks of same. We made considerable progress in compressing 'realistic' representations of neuronal computations into what we consider functionally equivalent input/output devices, which are now being incorporated into dynamic networks that learn associations and encode time. Our initial hypothesis about how to do this was rejected. Our new hypothesis offers great promise for scaling. This newer hypothesis resulted from examining simulations of 'realistic' neurons and thinking about the scaling problem. The latter was funded by the ONR.
Descriptors : *NEURAL NETS, *LEARNING MACHINES, COMPUTERIZED SIMULATION, COMPUTATIONS, EFFICIENCY, INPUT OUTPUT PROCESSING, NONLINEAR SYSTEMS, SYSTEMS ANALYSIS, DYNAMIC PROGRAMMING, SYNAPSE, CENTRAL NERVOUS SYSTEM, SYMPATHETIC NERVOUS SYSTEM, NEUROPHYSIOLOGY, SELF ORGANIZING SYSTEMS.
Subject Categories : Cybernetics
Distribution Statement : APPROVED FOR PUBLIC RELEASE