Optical neural net for matrix inversion
- 1 July 1989
- journal article
- Published by Optica Publishing Group in Applied Optics
- Vol. 28 (13) , 2499-2504
- https://doi.org/10.1364/ao.28.002499
Abstract
A novel optical neural net for performing matrix inversion is presented. This algorithm does not suffer from accumulative round-off error. Various optical architectures are introduced, and the circumstances in which each of them is advantageous are investigated.Keywords
This publication has 7 references indexed in Scilit:
- Computing with Neural Circuits: A ModelScience, 1986
- Correlation synthetic discriminant functionsApplied Optics, 1986
- Programmable real-time incoherent matrix multiplier for optical processingApplied Optics, 1986
- Optimization by Simulated AnnealingScience, 1983
- Fully-parallel relaxation algebraic operations for optical computersOptics Communications, 1982
- Acousto-optic signal processing: Convolution and correlationProceedings of the IEEE, 1981
- Stationary and nonstationary learning characteristics of the LMS adaptive filterProceedings of the IEEE, 1976