Probability Collectives with Collaborative Optimization (PCCO): A Novel Framework for Handling Complex Optimization Problems

Abstract

In this paper, a novel framework as a combination of Probability Collectives (PC) and Collaborative Optimization (CO) is proposed and detailed illustrated. The framework has a two-level structure which is similar to that of CO, but with the system level replaced by distributed PC based agents. This formulation maintains the advantage of CO while enhances the optimization and coordination ability at the system level. For better implementation, some adaption and improvement has been made to the origin PC method. The resultant PCCO framework shows satisfied performance in handling complex optimization problems with both efficiency and accuracy.