STRUCTURE AND CONVERGENCE PROPERTIES OF A RECURRENT NEURAL NETWORK

PDF

Authors
  1. Barton, S.A.
Corporate Authors
Defence Research Establishment Suffield, Ralston ALTA (CAN)
Abstract
The structure of a general form of recurrent neural network (RNN) that may be used for investigating pattern recognition and control capabilities is described. The details of C programs that configure, run and analyse the network are given. The convergence properties of the RNN, as a function of its structural and learning parameters, are investigated, and key conditions for stable, periodic, aperiodic and chaotic operation are established. The rate of convergence to a stable (fixed) state is shown to be strongly dependent on the learning parameters, and weakly dependent on the structural parameters. For stable systems, the potential for self-organized storage of repeatedly seen patterns (memory), the use of this in pattern recognition and object classification, and the potential for self-directed motor control in mobile autonomous machines are discussed.
Keywords
Image classification;Unmanned vehicles;Countermine operations
Report Number
DRES-M-1489 — Memorandum
Date of publication
01 Dec 1996
Number of Pages
73
DSTKIM No
97-01237
CANDIS No
501514
Format(s):
Hardcopy;Document Image stored on Optical Disk

Permanent link

Document 1 of 1

Date modified: