Tuning membership locations of a fuzzy classifier with ellipsoidal regions

We discuss a tuning method for a fuzzy classifier with ellipsoidal regions. Namely, we propose to tune the locations of membership functions. First, we transform the input data by the orthogonal matrix that diagonalizes the covariance matrix associated with a membership function. Then we tune the fuzzy rules, i.e., the locations of the membership functions, successively until there is no improvement in the recognition rate of the training data. We evaluate our method using four benchmark data sets and compare the performance of our method with that of tuning slopes of the membership functions