Boosting of granular models

In this study, we are concerned with the design of granular modeling being originally proposed by Pedrycz and Vasilakos. The enhancement of the development process comes in the form of the boosting mechanism applied to the generic model. In comparison with the original topology of the model studied so far, we augment it by a bias term and investigate its role in the overall architecture. Second, treating the granular model as a weak learner, we discuss the underlying mechanisms of boosting which in this setting has to be refined so that it handles the continuous case. From the design standpoint, we are interested in studying the following issues: (a) effectiveness of boosting when applied in this modeling framework along with its numeric quantification, and (b) impact of information granularity at which the granular models are developed on the improvements offered by boosting procedure itself. Numeric experiments help quantify the performance of the boosted granular models and gain a detailed view at the efficiency of the boosting strategy vis-à-vis different design scenarios.