Development of Control and Synchronization of Periodicity and Chaos in Recurrent Neural Networks for Robotic Applications

PDF

Authors
  1. Tam. Z.
  2. Ali, M.K.
  3. Barton, S.A.
Corporate Authors
Defence Research Establishment Suffield, Ralston ALTA (CAN);Lethbridge Univ, Lethbridge ALTA (CAN) Dept of Physics
Abstract
In this treatise, we present the results of our work that has been funded by Defence Research Establishment Suffield (DRES) since 1997 under R&D contract W7702-6-R607/A and the project title "Development of Control and Synchronization of Periodicity and Chaos in Recurrent Neural Networks (RNNs) for Robotic Applications". Standard RNNs involve nonchaotic dynamical systems with fixed points that serve for storage and retrieval of memories. This approach severely limits the socpe of applications of RNNs simply because nonchaotic RNNs constitute a small fraction of all possible RNNs. We embarked on treating RNNs from a much broader perspective and in the light of recent developments in dynamics. This required studying chaotic as ewll as nonchaotic RNNs. A brief description of our findings is given in this introduction.
Keywords
Machine intelligence;Machine training;Neural networks;Chaotic systems
Report Number
DRES-CR-1999-110 — Contract Report
Date of publication
01 Jul 1999
Number of Pages
186
DSTKIM No
CA000247
CANDIS No
512124
Format(s):
Hardcopy;Document Image stored on Optical Disk

Permanent link

Document 1 of 1

Date modified: