Biochip Technology Development
Abstract
Today the study of biological intelligence constitutes perhaps the most challenging yet rewarding frontier of research. A most compelling notion attracting many researchers is that such complex intelligence originates from a vast interconnected architecture containing approximately 1012 relatively simple computational units or neurons. The massively parallel architecture, sporting approximately 1015 adaptable neural interconnectors, is responsible for the extraordinary pattern processing capabilities of the brain. Among the pattern processing capabilities of interest to machine intelligence researchers pursuing engineering applications are fast pattern recognition by association, apparent infinite memory capacity, generalization, and robust, fault-tolerant performance. Furthermore, these system level capabilities are obtained biologically at modest expense, for the human brain consumes energy at a rate comparable to that of a single VLSI digital signal processing chip. Such efficiency arises from the rather conservative processing units, operating with 1 ms time constants and slowly communicating information (100ms) along passive neuronal channels.
Document Details
- Document Type
- Technical Report
- Publication Date
- Nov 09, 1990
- Accession Number
- ADA230564
Entities
People
- D. J. Ehrlich
- D. K. Astolfi
- M. D. Eggers
- S. G. Cann
- Y. V. Grinbush
Organizations
- Massachusetts Institute of Technology