In contrast to the previously examined algorithms, the algorithms in this category assume that each data vector may belong to (or may be compatible with) more than one cluster up to a certain number.