Data mining with cellular discrete event modeling and simulation

Data mining is the process of extracting patterns from data. A main step in this process is referred to as data classification. In this work, we investigate the use of the Cell-DEVS formalism for classifying data. The cells in a Cell-DEVS based grid are individually very simple but together they can represent complex behavior and are capable of self-organization. Three classifier models are implemented using Cell-DEVS. Different simulation scenarios are presented investigating the effect of Von Neumann versus Moore neighborhood in the classifiers’ models. We show that effective classification performance, comparable to those produced by complex data mining techniques, can be obtained from the collective behavior of discrete-event cellular grids.