M-FMCN: Modified Fuzzy Min-Max Classifier Using Compensatory Neurons

摘要

A new fuzzy Min-Max classifier is proposed that uses modified compensatory neurons. The proposed classifier is online, single-pass and supervised method that is based on fuzzy Min-Max neural network classifier with compensatory neurons. In this method for handling overlapping regions that mainly are created in borders, a modified compensatory nod with a radios-based transition function is used which increases the classification accuracy in discriminating cases. On contract of modifications in the structure of the algorithm, time and space complexity of the algorithm has been decreased and experimental results show that the proposed method is less sensitive to external parameters that are provided by user.