New results on diagnosis by fuzzy pattern recognition

We use the classification method Fuzzy Pattern Matching (FPM) to realize the industrial and medical diagnosis. FPM is marginal, i.e., its global decision is based on the selection of one of the intermediate decisions. Each intermediate decision is based on one attribute. Thus, FPM does not take into account the correlation between attributes. Additionally, FPM considers the shape of classes as convex one. Finally the classes are considered as equi-important by FPM. These drawbacks make FPM unusable for many real world applications. In this paper, we propose improving FPM to solve these drawbacks. Several synthetic and real data sets are used to show the performances of the Improved FPM (IFPM) with respect to classical one as well as to the well known classification method K Nearest Neighbours (KNN). KNN is known to be preferment in the case of data represented by correlated attributes or by classes with different a priori probabilities and non convex shape.