A support vector machine-based fuzzy classification technique

Abstract

A novel type of learning machine based on statistical learning theory is the support vector machine (SVM). SVMs have been widely employed in classification, regression, and pattern recognition because of their strong generalisation capabilities. This research proposes a novel fuzzy classification method (FCM) based on SVM for data containing numerical condition attributes and decision attributes. This process first confuses some classes (linguistic words)' choice attributes before training the decision function (classifier). The decision function provides the matching class and its membership degree as a fuzzy decision for a fresh sample rather than predicting the value of its decision characteristic. In contrast to crisp decisions, this consequence of the fuzzy decision is more objective and intuitive to comprehend. The classification algorithm is provided, as well as the design principle.