We introduce a new approach to the problem of
overlapping clustering. The main idea is to formulate overlapping
clustering as an optimization problem in which each data point
is mapped to a small set of labels, representing membership to
different clusters.