Automatic Extraction of Linguistic Description from Fuzzy Rule Base
This work addresses the need for more interpretable AI outputs in explainable artificial intelligence (XAI) by automating linguistic descriptions for domain experts.
The paper tackled the problem of automatically generating natural English language descriptions from fuzzy rule bases in neuro-fuzzy systems, resulting in a fully implemented and publicly available tool for this extraction.
Neuro-fuzzy systems are a technique of explainable artificial intelligence (XAI). They elaborate knowledge models as a set of fuzzy rules. Fuzzy sets are crucial components of fuzzy rules. They are used to model linguistic terms. In this paper, we present an automatic extraction of fuzzy rules in the natural English language. Full implementation is available free from a public repository.