ABSTRACT
This chapter reviews algorithms for generating alternative clusterings, which is one of the prime tasks in the field of multiple clustering analysis. It highlights connections to the areas of multiview clustering and subspace clustering, which are distinct, yet closely related. In multiview clustering, the aim is to learn a single clustering using multiple sources of the data. The chapter also reviews the different dimensions that may be used for assessing the behaviour of alternative clustering algorithms (ACA). ACAs may be characterized in a range of different ways. Naive generation is a very common technique employed by users who are not familiar with alternative clustering. An extension of the naive technique is the approach of meta clustering. The COALA method takes as input a similarity matrix and a single existing clustering as background knowledge. Another approach to the generation of alternative clusterings is based on the use of objective functions using information theoretic principles.
