Discriminative Clustering in Fisher Metrics

Discriminative clustering (DC) finds a Voronoi partitioning of a primary data space that, while consisting of local partitions, simultaneously maximizes information about auxiliary data categories. DC is useful in exploration and in finding more coarse or refined versions of already existing categori es. Theoretical results suggest that Voronoi partitions in the socalled Fisher metric would outperform partitions in the Euclidean metric. Here we use a local quadratic approximation of the Fi sher metric, derived from a conditional density estimator, in defining the partitions and show that the resulting algorithms outperform the conventional variants.