Evolutionary Machine Learning in Linguistic Knowledge Extraction

This book is proposing a hybrid algorithm of two fuzzy genetic-based machine learning approaches - Michigan and Pittsburgh - for designing fuzzy rule-based classification systems. The search ability of each approach is examined to efficiently find fuzzy rule-based systems with high classification accuracy. These two approaches are combined into a single hybrid algorithm. The generalization ability of fuzzy rule-based classification systems designed by the proposed hybrid algorithm is examined on real data sets. Experimental results show that the hybrid algorithm has higher search ability within a population of individual rules and within a population of rule sets.