Year

1996

Degree Name

Master of Computer Science

Department

Department of Computer Science

Abstract

This thesis introduces a dynamic recognition neural network model (DRNNM) that provides a theoretical basis for the resolution of a number of pattern recognition problems. These problems consist of: The Binding Problem; The Correspondence Problem; The Learning Complexity Problem and The Knowledge Transference and Extension Problem. The Thesis also addresses related issues, such as: recognition with scarce training resources; dynamic feature extraction and a methodology for reducing learning conflict or crosstalk.

Share

COinS
 

Unless otherwise indicated, the views expressed in this thesis are those of the author and do not necessarily represent the views of the University of Wollongong.