THE DESIGN OF AN INDEPENDENT,OPERATIONAL, DECISION MAKING CONTROL UNIT FOR AUTOMATIC RECOGNITION

IT IS PROPOSED TO INVESTIGATE, BY A SERIES OF EXPERIMENTS THE DESIGN OF A STORAGE AND CONTROL MECHANISM CALLED A RULING. IS A MATHEMATICAL CONCEPT WHICH LENDS ITSELD TO IMPLEMENTATION ON A CONVENTIONAL COMPUTER. IT IS AN ENVIRONMENT WHICH PERMITS LEARNING IN A REAL-TIME, UNKNOWN ENVIRONMENT, WITH CONTROL BASED ON PAST AND PRESENT LEARNED EXPERIENCE. IN A TYPICAL ENVIRONMENT, A RULING CNSUMING 10 TO THE ELEVENTH POWER BITS IS EQUIVALENT TO A DATA BASE OF 10 TO THE 54TH POWER BITS. UTILIZING A RULING, VAST AMOUNTS OF INFORMATION CAN BE INCORPORATED INTO A RELATIVELY SMALL SYSTEM, PERMITTING THE PERFORMANCE OF ACTIVITIES BASED ON THAT ENORMOUS EXPERIENCE POOL. THIS EFFORT \WILL INVOLVE: 1) IMPLEMENTING A RULING ON A COMPUTER, 2) SELECTING AN APPLICATION AREA, 3) DESCRIBING AN ALPHABET REPRESENTING THE ACTIVITIES OF THE APPLICATION, AND THEN 4) BUILDING THE EXPERIENCE POOL. ONCE THE EXPERIENCE POOL IS LARGE ENOUGH TO PERMIT THE CONTROL FUNCTION TO OPERATE, A NEW SET OF UNKNOWN EVENTS WILL BE SUPPLIED. MEASURES OF PERFORMANCE WILL BE GIVEN TO DESCRIBE THE PERFORMANCE IN BOTH THE "LEARNED" AND "UNLEARNED" ENVIRONMENTS. THE OBJECTIVE OF THIS EFFORT IS DEMONSTRATE THE APPLICABILITY OF THE PROCESS OF ACCEPTING AND "CONSOLIDATING" INFORMATION INTO A RULING AND THEN THAT INFORMATION TO MAKE DECISIONS BASED ON THAT EXPERIENCE. THIS DEMONSTRATION WILL BE ACCOMPISHED BY PERFORMING THE EXPERIMENTS IDENTIFIED IN THIS PROPOSAL.