A clustering-based fuzzy classifier

In this work we propose to use fuzzy prototypes for classification tasks We create the prototypes by first clusterizing separately the available data for each class Then we create a fuzzy set around each class center according to clusterization indices that take into account both local and global data We also investigate the use of an index to guide the process of obtaining a suitable number of clusters for posterior classification We illustrate the approach using a simple example and show some results for some well known data sets