Comparing Fuzzy Clusterings in High Dimensionality

Due to the specificity of clustering, a problem that is intrinsically ill-posed, there are several approaches to comparing clusterings. Comparison of clusterings obtained in different conditions is often the only affordable evaluation strategy, due to the lack of a ground truth. In this chapter we address a class of dimensionality-independent methods which can be applied in the presence of a high-dimensional input space. Specifically, we review some generalizations of this class of methods to the case of fuzzy clustering, in several variants.