Induction of Fuzzy Prototypes with Feature Selection

In the sequel, we shall be concerned with the problem of automated learning from a set of data. We propose an algorithm, inspired by hierarchical clustering techniques, that finds groupings of similar objects and represents them using fuzzy prototypes. It also learns weights for the attributes describing the data, allowing us to perform feature selection. The potential of the resulting method is illustrated by its application to an unsupervised learning problem.