Learning Fuzzy Rule Based Classifier with Rule Weights Optimization and Structure Selection by a Genetic Algorithm

This paper presents a method for designing fuzzy rule based systems for pattern recognition. The resulting model is interpretable as linguistic rules and can be used for deep understanding of data. The classifier performance is optimized in the least squares sense and the model complexity is minimized in a structure selection search, performed by a genetic algorithm The method is tested against benchmark classification problems found in the literature, with good results.