A review on evolving interval and Fuzzy granular systems

Abstract

This article provides definitions and principles of granular computing and discusses the generation and online adaptation of rule-based models from data streams. Essential notions of interval analysis and fuzzy sets are addressed from the granular computing point of view. The article also covers different types of aggregation operators which perform information fusion by gathering large volumes of dissimilar information into a more compact form. We briefly summarize the main historical landmarks of evolving intelligent systems leading to the state of the art. Evolving granular systems extend evolving intelligent systems allowing data, variables and parameters to be granules (intervals and fuzzy sets). The aim of the evolution of granular systems is to fit the information carried by data streams from time-varying processes into rule-based models and, at the same time, provide granular approximation of functions and linguistic description of the system behavior