Accession Number : ADP008236

Title :   Persistent Spectral Hole Burning Applications for Massive Optical Neural Network Computers,

Corporate Author : SPARTA INC LEXINGTON MA

Personal Author(s) : Henshaw, Philip D. ; Lis, Steven A.

Report Date : 22 MAY 1992

Pagination or Media Count : 4

Abstract : Neural networks require two types of operations; interconnections, which define how the output of one state affects the input of the next, and non-linear operations, which relate the inputs of a state to its output. Interconnections, which require many signals passing through the same space, are best performed with photons, which do not interact with one another. Non-linear operations require interaction (i.e., cross products) between the various inputs to a state, and are best performed with electrons, which interact strongly through their electrical charge. In a typical neural network architecture, almost all of the computation required is associated with the interconnections, and only a tiny fraction is associated with the non-linear operations (sigmoidal response or thresholding) performed at each state. In this paper we will present an architecture which uses both photons and electrons in a natural manner to perform all the functions required for a complete neural network architecture. A schematic of this architecture is shown in Figure 1. Almost all of the computations are performed optically in parallel, providing the capability to implement extremely large neural networks.

Descriptors :   *NEURAL NETS, *COMPUTER ARCHITECTURE, *HOLOGRAMS, *OPTICAL CIRCUITS, *OPTICAL STORAGE, *MOLECULAR SPECTROSCOPY, ELECTRONS, INPUT, INTERACTIONS, NETWORKS, OUTPUT, PHOTONS, RESPONSE, SIGNALS, SIGNAL PROCESSING, THREE DIMENSIONAL, FREQUENCY, FOUR DIMENSIONAL, CHIPS(ELECTRONICS).

Subject Categories : Cybernetics
      Atomic and Molecular Physics and Spectroscopy

Distribution Statement : APPROVED FOR PUBLIC RELEASE