Fault tolerance of opto-electronic neural networks

Anjan K. Ghosh

doi:10.1117/12.2321724

1 December 1991 Fault tolerance of opto-electronic neural networks

Anjan K. Ghosh

Proceedings Volume 1563, Optical Enhancements to Computing Technology; (1991) https://doi.org/10.1117/12.2321724
Event: San Diego, '91, 1991, San Diego, CA, United States

Abstract

Many ideas on opto-electronic realizations of various neural networks have been described over the last few years. Most of them use analog optical devices and have been shown to work satisfactorily as simple content addressable memories. An important question is how much of the faults and defects in the storage and retrieval systems, that is, the spatial and temporal errors in the spatial light modulators or the photorefractive arrays and the detector arrays these neural network realizations can tolerate. One measure of fault tolerance is the probability of correct retrieval or classification when the network is used as a content addressable memory. Using recent theories we will derive upper bounds to the probability of errors in various popular opto-electronic neural network realizations as a function of the size, capacity, and the magnitude of spatial and temporal errors. Guidelines on the design and implementations of opto-electronic neural networks will be derived.

Citation Download Citation

Anjan K. Ghosh "Fault tolerance of opto-electronic neural networks", Proc. SPIE 1563, Optical Enhancements to Computing Technology, (1 December 1991); https://doi.org/10.1117/12.2321724

ACCESS THE FULL ARTICLE

INSTITUTIONAL
Select your institution to access the SPIE Digital Library.

SELECT YOUR INSTITUTION

PERSONAL
Sign in with your SPIE account to access your personal subscriptions or to use specific features such as save to my library, sign up for alerts, save searches, etc.

PERSONAL SIGN IN

No SPIE Account? Create one

PURCHASE THIS CONTENT

SUBSCRIBE TO DIGITAL LIBRARY

50 downloads per 1-year subscription

Members: $195

Non-members: $335 ADD TO CART

25 downloads per 1 - year subscription

Members: $145

Non-members: $250 ADD TO CART

PURCHASE SINGLE ARTICLE

Includes PDF, HTML & Video, when available