Expand this Topic clickable element to expand a topic
Skip to content
Optica Publishing Group

Optoelectronic learning machine

Not Accessible

Your library or personal account may give you access

Abstract

Adaptive resonance is a theory of unsupervised learning in neural networks.1 We propose an invention that allows parallel execution of the most computationally intensive parts of the architecture by using optical hardware. The rest of the design is carried out in electronics. The basic configuration of the electro-optical ART unit is a two-lens optical correlator. A spatial light modulatoris configured as a binary phase-only filter and contains the two-dimensional Fourier transform of the input pattern. The plane of this device is between the two lenses. The template plane is to the left of the first lens. It contains multiple templates simultaneously. The lenses are chosen such that the paraxial approximation applies, and the template Fourier transforms will overlap. On the output plane, we have a charge-coupled device (CCD) camera. This plane will have correlation peaks corresponding to templates that closely match the input image. All electronic calculations and control of the spatial light modulators are done by a microVAX. This is also used to process the CCD camera output via a frame grabber.

© 1990 Optical Society of America

PDF Article
More Like This
A Stochastic Optoelectronic Learning Machine

Nabil H. Farhat and Zon Yin Shae
PDP13 OSA Annual Meeting (FIO) 1988

Optoelectronic connectionist machines

L. Zhang and K. M. Johnson
MUU4 OSA Annual Meeting (FIO) 1991

Multimode fiber projection with machine learning

Babak Rahmani, Damien Loterie, Eirini Kakkava, Navid Borhani, Ugur Tegin, Demetri Psaltis, and Christophe Moser
HTu5B.3 Digital Holography and Three-Dimensional Imaging (DH) 2020

Select as filters


Select Topics Cancel
© Copyright 2024 | Optica Publishing Group. All rights reserved, including rights for text and data mining and training of artificial technologies or similar technologies.