PASCAL - Pattern Analysis, Statistical Modelling and Computational Learning

A Dependence Maximization View of Clustering
Le Song, Alex Smola, Arthur Gretton and karsten borgwardt
In: ICML 2007, 20 June - June 24 2007, Oregon USA.


We propose a family of clustering algorithms based on the maximization of dependence between the input variables and their cluster labels, as expressed by the Hilbert-Schmidt Independence Criterion (HSIC). Under this framework, we unify the geometric, spectral,and statistical dependence views of clustering, and subsume many existing algorithms as special cases (e.g. k-means and spectral clustering). Distinctive to our framework is that kernels can also be applied on the labels, which can endow them with particular structures. We also obtain a perturbation bound on the change in k-means clustering.

EPrint Type:Conference or Workshop Item (Paper)
Project Keyword:Project Keyword UNSPECIFIED
Subjects:Computational, Information-Theoretic Learning with Statistics
Theory & Algorithms
ID Code:3132
Deposited By:Arthur Gretton
Deposited On:21 December 2007