Semi-Supervised Constrained Clustering with Cluster-Dependent Constraints Seminar

In this work we present a heuristic procedure to perform clustering subject to constraints for K classes, when the constraints refer to the cluster themselves, that is, an element must satisfy a set of constraints given the cluster in which the element is in. These constraints can link clusters, or can just restrict the space of feasible points that the cluster can permit. The problem, combinatorial in nature, appears when incorporating external knowledge about the structure of the clusters in segmentation problem, and is of use in Business Analytics applications, as will be shown in the presentation.