EXPERT SYSTEM FOR AUTOMATIC DECISION MAKING IN TARGET DETECTION AND CLASSIFICATION

A SYSTEM IS PROPOSED FOR UTILIZING EXPERT KNOWLEDGE TO AUTOMATICALLY GENERATE DECISION TREES WITH APPLICATIONS IN TARGET DETECTION AND CLASSIFICATION. THE PROPOSED WORK REPRESENTS THE FIRST FORMALIZED ATTEMPT TO COMBINE EXPERT SYSTEM TECHNOLOGY AND AUTONOMOUS DECISION MAKING FOR TASKS SUCH AS COMPUTER VISION, PATTERN RECOGNITION, AND INTELLIGENT CONTROL. IN THE PROPOSED SYSTEM, A KNOWLEDGE BASE CONSISTING OF A SET OF PRODUCTION RULES IS CREATED BY THE USER. FROM THE KNOWLEDGE BASE, THE SYSTEM CAN THEN AUTOMATICALLY CREATE A DECISION TREE REPRESENTATION OF THE SOLUTION TO THE PROBLEM THAT SATISFIES A SPECIFIED OPTIMALITY CRITERION. FROM THAT OPTIMAL REPRESENTATION OF THE TREE, A HIGH LEVEL LANGUAGE IMPLEMENTATION IS THEN GENERATED. IF THE KNOWLEDGE BASE IS MODIFIED, THE SYSTEM PRODUCES A NEW DECISION TREE THAT INCORPORATES ALL THE CHANGES. THIS MEANS THAT THE USER DOES NOT HAVE TO WORRY ABOUT THE ACTUAL GENERATION, IMPLEMENTATION AND VISUALIZATION OF THE TREE. THE RESULT WILL BE A PROBLEMSOLVING APPROACH WHERE CNCEPTUAL AND EXPERT KNOWLEDGE, RATHER THAN HEURISTIC DETAILS, WILL BE THE DRIVING FORCE. THE PROBABILITY OF SUCCESS OF PHASE I IS HIGH BECAUSE MOST OF THE BACKGROUND WORK FOR DEMONSTRATING FEASIBILITY HAS BEEN DONE AND PARTS OF THE SYSTEM HAVE ALREADY BEEN IMPLEMENTED.