A fuzzy classifier for imbalanced and noisy data

This paper deals with concept learning in the presence of noise (overlap) and imbalance in the training set. The starting assumption is that recognition of the smaller class is much more important than that of the larger class. A fuzzy classifier capable to achieve this, based on the relation between fuzzy sets and probability distributions as mediated by the theory of mass assignment is presented. Two approaches to construct fuzzy sets - basic and modified - using the lpd and mpd selection rules are investigated. Preliminary results suggest that use of the mpd selection rule in conjunction with the modified approach is better for recall of the small class at a small cost to the recognition of the negative class.