Fuzzy Classifier with Probabilistic IF-THEN Rules

摘要

The typical fuzzy classifier consists of rules each one describing one of the classes. This paper presents a new fuzzy classifier with probabilistic IF-THEN rules. A learning algorithm based on the gradient descent method is proposed to identify the probabilistic IF-THEN rules from the training data set. This new fuzzy classifier is finally applied to the well-known Wisconsin breast cancer classification problem, and a compact, interpretable and accurate probabilistic IF-THEN rule base is achieved.