Contrast classification rules for mining local differences in medical data


In this paper, we formally prove that the classification rules formed on the basis of contrast patterns are guaranteed to be of a high quality. We propose to use the new ‘Sets of Contrasting Rules’ pattern for the identification of local differences between the classes of the dataset. Being essentially a contrast pattern formed of several classification… (More)
DOI: 10.1109/IDAACS.2017.8095213


1 Figure or Table