A Novel Framework for Enhanced Interpretability in Fuzzy Cognitive Maps

• DOI: 10.36227/techrxiv.22718032.v1
• Publication Date: 2023-05-02T15:02:57Z
• Authors: Marios Tyrovolas; X. San Liang; Chrysostomos Stylios
• Keywords: Interpretability; Fuzzy cognitive map; Spurious relationship; Representation; Causal reasoning

<p>A Fuzzy Cognitive Map (FCM) is a graph-based tool for knowledge representation that intends to model any complex system through an interactive structure of nodes interacting with each other causal relationships. Owing their flexibility and inherent interpretability, FCMs have been used in various modeling prediction tasks, particularly situations where humans make final decisions, such as industrial anomaly detection. However, can unintentionally absorb spurious correlations presented collected data during development, leading poor accuracy interpretability. To address this limitation, article proposes novel framework constructing based on the Liang-Kleeman Information Flow (L-K IF) analysis, inference tool. The actual relationships are identified from using automatic search algorithm, these then imposed constraints FCM learning procedure rule out improve predictive explanatory power model. Numerical simulations were conducted by comparing proposed approach state-of-the-art FCM-based models, thereby demonstrating promising performance developed FCM.</p>