Abstract
The emerging category of evolvable fuzzy systems has opened a new uncharted territory of system modeling by enhancing the capabilities of existing fuzzy models and formulating new methodological and algorithmic challenges and opportunities. In this study, we revisit the underlying concept and identify a number of essential optimization problems arising therein. It is shown that the behavior and characteristics of evolvable systems can be classified under the rubric of perception-based evolvability (being inherently associated with the human-centric systems and the development of efficient mechanisms of relevance feedback) and a distribution of knowledge representation resources of evolvable systems. We elaborate on the essence of these problems and define the corresponding optimization criteria. A selected detailed design scenario is presented as well in which the dynamics of information granules is exploited as a vehicle to cope with the evolving modeling environment.