A Survey of Approaches to Context Adaptation of Fuzzy Systems

Thanks to their modeling and interpretability capabilities, Mamdani-type fuzzy rule-based systems (FRBSs) have been widely employed in several application domains. Often, the design of an FRBS involves two phases: first, universal rules are derived from a high-level and context-free knowledge, then these rules are instantiated to a specific context so as to generate a working FRBS. In this paper, we review existing approaches to context adaptation of FRBSs. We develop a uniform notation and a common framework for these approaches, and we highlight open issues and future research topics.