Jordan University of Science and Technology
Knowledge-Based Constrained Function Optimization Using Cultural Algorithms with an Enhanced Social Influence Metaphor
Authors:
Mostafa Ali, Robert Reynolds, Rose Ali, and Ayad Salhieh
Abstract:
In this research we present a new framework based on Cultural
Algorithms using an enhanced social fabric influence function to solve
nonlinearly constrained global optimization problems. We identify how
knowledge sources used by Cultural Algorithms are combined to direct the
decisions of the individual agents during the problem solving process using an
influence function family based upon a Social Fabric metaphor. Guided
interactions between the population swarms and these knowledge sources
produced emergent phases of problem solving. This implies that the social
interaction of individuals coupled with their interaction with a culture within
which they are embedded provides a powerful vehicle for the solution of these
problems. Results demonstrate that this approach can successfully extract
interesting emergent patterns in the Belief space and improve the search
efficiency by avoiding local Optima, and converge to an approximate global
minimizer asymptotically. Different parameter combinations can affect the rate
of solution.