الفهرس 
AcknowledgementOnly 14 pages are availabe for public view
 |  | from | 174 |  |  |
| | | from | 174 | | |
Abstract
work presented in this thesis is to demonstrate how | neural networks can be used, in addition to classical knowledge representation tech-iques, in a distributed and collaborative manner to approach human performance dealing with complex tasks that contain pattern association and recognition. A of the literature regarding expert systems, blackboard architectures, connec-list models, and distributed systems is first presented. The proposed architecture, blackboard connectionist architecture, is then presented to guide the develop-it of connectionist expert systems and followed by a simulated, multi-tasking Dtotype implementation. Finally, two real-world applications from two different jlication domains, an Arabic text recognizer and a prostate cancer patients’ clas-er, have been considered and developed under the prototype