Extraction of Fuzzy Rules by Using Support Vector Machines
Shuwei Chen, Jie Wang, Dongshu Wang · 2008
This paper proposes an architecture to extract fuzzy rules based on support vector machines (SVMs). Firstly, support vectors are obtained from the training data set to generate fuzzy if-then rules with membership functions described in terms of kernel functions via support vector machine learning procedure. Then, a combined fuzzy rule base is created based on both the generated rules and linguistic rules of human experts. Thus, it has the inherent advantages that the rule base is optimized automatically during the SVM learning procedure, and, takes both "subjective" experts' prior knowledge and "objective"' training data into account. An example is given to show the effectiveness of the proposed method.