Optical Relational-Graph Rule-Based Processor for Structural-Attribute Knowledge Bases,
CARNEGIE-MELLON UNIV PITTSBURGH PA DEPT OF ELECTRICAL AND COMPUTER ENGINEERING
Pagination or Media Count:
An optical relational-graph or decision-net processor is advanced, and one optical system design for such a processor is provided. The general description of this processor is provided as well as two designs for a specific case study. A rule-based design of a knowledge base of facts using structural rather than functional attributes of the object is detailed. A general technique is advanced to organize a fact-based knowledge base into a rule-based knowledge base using structural rather than functional attributes. A multiclass 3-D distorted object identification and classification problem is considered as our case study. A general relational-graph design methodology is advanced, and specific designs for our case study are then presented. Multidecision and binary relational-graph designs are advanced, and impressive initial simulation results with each are noted. Reprint.