An Effective Multiresolution Hierarchical Granular Representation Based Classifier Using General Fuzzy Min-Max Neural Network

Motivated by the practical demands for simplification of data toward being consistent with human thinking and problem-solving, as well tolerance uncertainty, information granules are becoming important entities in processing at different levels abstraction. This article proposes a method to construct classifiers from multiresolution hierarchical granular representations using hyperbox fuzzy sets. The proposed approach forms series inferences hierarchically through many An attractive characteristic our classifier is that it can maintain high accuracy comparison other min-max models low degree granularity based on reusing knowledge learned lower In addition, reduce size significantly handle uncertainty incompleteness associated real-world applications. construction process consists two phases. first phase formulate model greatest level granularity, while later stage aims complexity constructed deduce higher abstraction levels. Experimental analyses conducted comprehensively both synthetic real datasets indicated efficiency terms training time predictive performance types neural networks common machine learning algorithms.