DECISION-THEORETIC CONTROL OF ARTIFICIAL INTELLIGENCE SCHEDULING SYSTEMS

A PROTOTYPE DECISION-THEORETIC SCHEDULER (DTS) WILL BE DEVELOPED THAT EMPLOYS STATE-OF-THE-ART PROBABILISTIC INFERENCE TECHNOLOGY TO EXPEDITE THE SEARCH FOR EFFICIENT SOLUTIONS TO CONSTRAINT-SATISFACTION PROBLEMS. THE APPROACHINVOLVES ASSESSING THE PERFORMANCE OF HEURISTIC CONTROL STRATEGIES, WHICH ARE NORMALLY HARD-CODED INTO SCHEDULING SYSTEMS, AND USING PROBABILISTIC INFERENCE TO AGGREGATE THIS INFORMATION IN THE LIGHT OF A GIVEN PROBLEM'S FEATURES. THE COMPANY RECENTLY INTRODUCED A SIMILAR APPROACH TO SOLVING SINGLE-AGENT GRAPH SEARCH PROBLEMS, YIELDING ORDERS-OF-MAGNITUDE IMPROVEMENT OVER TRADITIONAL SEARCH ALGORITHMS ON A CLASSICAL TEST DOMAIN. RECENT WORK IN THE CONSTRAINT-SATISFACTION COMMUNITY SUGGESTS THAT THIS ADVANTAGE MAY BE EVEN MORE DRAMATIC WHEN APPLIED TO TYPICAL SCHEDULING PROBLEMS. IN ADDITION, DTS PROMISES TO PROVIDE A RICHER FRAMEWORK FOR HUMAN SCHEDULING EXPERTS TO EXPRESS DECLARATIVELY CONFLICTING MULTI-ATTRIBUTE OBJECTIVES THAT THEIR SCHEDULES MUST SATISFY. A PROTOTYPE DECISION-THEORETIC SCHEDULER (DTS) WILL BE DEVELOPED THAT EMPLOYS STATE-OF-THE-ART PROBABILISTIC INFERENCE TECHNOLOGY TO EXPEDITE THE SEARCH FOR EFFICIENT SOLUTIONS TO CONSTRAINT-SATISFACTION PROBLEMS. THE APPROACHINVOLVES ASSESSING THE PERFORMANCE OF HEURISTIC CONTROL STRATEGIES, WHICH ARE NORMALLY HARD-CODED INTO SCHEDULING SYSTEMS, AND USING PROBABILISTIC INFERENCE TO AGGREGATE THIS INFORMATION IN THE LIGHT OF A GIVEN PROBLEM'S FEATURES. THE COMPANY RECENTLY INTRODUCED A SIMILAR APPROACH TO SOLVING SINGLE-AGENT GRAPH SEARCH PROBLEMS, YIELDING ORDERS-OF-MAGNITUDE IMPROVEMENT OVER TRADITIONAL SEARCH ALGORITHMS ON A CLASSICAL TEST DOMAIN. RECENT WORK IN THE CONSTRAINT-SATISFACTION COMMUNITY SUGGESTS THAT THIS ADVANTAGE MAY BE EVEN MORE DRAMATIC WHEN APPLIED TO TYPICAL SCHEDULING PROBLEMS. IN ADDITION, DTS PROMISES TO PROVIDE A RICHER FRAMEWORK FOR HUMAN SCHEDULING EXPERTS TO EXPRESS DECLARATIVELY CONFLICTING MULTI-ATTRIBUTE OBJECTIVES THAT THEIR SCHEDULES MUST SATISFY.