Integrating Fuzzy Cognitive Maps with Statistical Reasoning in Explainable Artificial Intelligence.

Fuzzy cognitive maps and statistical reasoning are considered from the standpoint of explainable artificial intelligence. This approach means that today the artificial intelligence models should also be transparent, conceivable and have explanatory power for the users. We will consider how fuzzy cognitive maps may respond to these challenges because these maps provide a simple and conceivable method for modelling complex phenomena of the real world. Since their construction often stems from sophisticated machine learning and metaheuristic optimization methods as well as their concept values and interrelationships may base on subjective and ambiguous interpretations, we are still encountering problems in this sense. Our approach will apply statistical reasoning and theories, especially regression models, to fuzzy cognitive maps. Hence, we may avoid complex mathematical calculations and operate with objective, unambiguous and conceivable models, and in general, achieve better the aims of explainable artificial intelligence. [ABSTRACT FROM AUTHOR]