A hierarchical fuzzy rule-based learning system based on an information theoretic

This paper proposes a new novel method for the online construction of a Hierarchical Fuzzy Rule Based System (FRBS) to accurately model a function while retaining a level of human interpretability. The algorithm uses an information theoretic approach to limit the amount of uncertainty within each decision and to determine when a rule does not effectively model the underlying decision space. Experimental results are provided which compare the performance of the proposed system with existing approaches.

• Type: Conference
• Year: 2003
• Where: EUSFLAT
• Authors: Antony Waldock, Brian Carse, Chris Melhuish
• Keywords: EUSFLAT 2003, EUSFLAT 2007, Hierarchical Fuzzy Rule, Human Interpretability, Information Theoretic Approach