Large Minded Reasoners for Soft and Hard Cluster Validation -- Some Directions

Abstract

In recent research, validation methods for soft and hard clustering through general granular rough clusters are proposed by the first author. Large-minded reasoners are introduced and studied in the context of new concepts of nonstochastic rough randomness in a separate paper by her. In this research, the methodologies are reviewed and new lowcost scalable methodologies and algorithms are invented for computing granular rough approximations of soft clusters for many classes of partially ordered datasets. Specifically, these are applicable to datasets in which attribute values are numeric, vector valued, lattice-ordered or partially ordered. Additionally, new research directions are indicated.