New experimental results in fuzzy clustering

Abstract This paper is mainly devoted to the description of experiments using improved techniques for clustering data in fuzzy sets. Previous methods provided good fuzzy dichotomies but failed for partitions in a larger number of sets. The modifications presented here resulted in good fuzzy classifications in any, previously established, number of clusters. The modifications are described both theoretically and computationally. The effect of variations in the function that maps the original distance structure in the data space and that prescribed for the classification space is examined. Numerous computational examples are presented.