A granular computing approach to improve large attributes learning

Based on the concept of granular computing, this article proposes a novel Boolean conversion (BC) method to reduce data attribute number for the purpose of improving the efficiency of learning in artificial intelligence. Data with large amount of attributes usually cause a system freezes or shuts down. The proposed method combines large amount attributes to smaller number ones by the way of Boolean method. Three data sets are used to compare the learning accuracies and efficiencies by Bayesian networks (BN), C4.5 decision tree, support vector machine (SVM), artificial neural network (ANN), fuzzy neural network (FNN, neuro-fuzzy), and mega-fuzzification learning methods. Results indicate that the proposed BC method can improve the efficiency of machine learning and the accuracy is not worse.