MVLSC Home · Issue Contents · Forthcoming Papers
A Semi-Fuzzy Approach to Fuzzy Cognitive Maps by Applying Explainable Artificial Intelligence
Vesa A. Niskanen
Fuzzy cognitive maps are studied from the standpoints of statistics and explainable artificial intelligence. The models of the prevailing fuzzy cognitive maps already meet well the challenges of the explainable artificial intelligence, but their outcomes are often interpreted and estimated with too subjective and ambiguous assessments. A semi-fuzzy method is suggested for resolving these problems. In this approach, statistical methods, especially linear regression models and fuzzy rule-based systems were used to fuzzy cognitive map construction. Two practical examples were also provided for justifying this method. We noticed that, when applying statistical reasoning, more objective, unambiguous, and conceivable models may be constructed. This approach also met the challenges provided by the explainable artificial intelligence.
Keywords: Fuzzy cognitive maps, regression analysis, explainable artificial intelligence