Optical Neural Nets for Scene Analysis
Abstract
A hybrid optical/digital neural net for scene an analysis described. It combines pattern recognition and neural net techniques. New algorithms, architectures and applications are described for optimization neural nets (a mixture neural net for image spectrometry, cubic and quadratic neural nets for multitarget tracking, and a matrix inversion neural net), production system neural nets, symbolic neural nets and a new adaptive learning neural net (the adaptive clustering neural net). Progress in the first six months on these seven neural nets and our hardware are presented. Keywords: Artificial intelligence, Computer architecture, Networks, Data bases, Data processing. (AW)
Document Details
- Document Type
- Technical Report
- Publication Date
- Oct 23, 1989
- Accession Number
- ADA213986
Entities
People
- David Cassent
Organizations
- Carnegie Mellon University