A hierarchical genetic fuzzy rule-based classifier for high-dimensional classification problems
Dimitris G. Stavrakoudis, Ioannis Z. Gitas, John B. Theocharis · 2011
This paper proposes a novel Hierarchical Genetic Fuzzy Rule-Based Classification System (HGΓRBCS), targeted at effectively handling high-dimensional classification tasks. A hierarchical fuzzy rule base comprises rules with linguistic terms from a multi-granular fuzzy sets database, whereby lower levels define thicker granularities of the input space fuzzy partition. The proposed system is developed through sequential repeating steps: in each step a fuzzy rule base is created using a given granularity. Subsequently, the best performing rules are inserted in the hierarchical rule base and the process is repeated again, considering a thicker granularity. The whole process is coordinated by a boosting scheme, which localizes new rules in uncovered regions of the feature space. Comparative results for various real-world high-dimensional classification problems indicate the effectiveness of the proposed methodology.