Information theoretic clustering using minimal spanning trees Conference Paper


Author(s): Müller, Andreas; Nowozin, Sebastian; Lampert, Christoph
Title: Information theoretic clustering using minimal spanning trees
Title Series: LNCS
Affiliation IST Austria
Abstract: In this work we propose a new information-theoretic clustering algorithm that infers cluster memberships by direct optimization of a non-parametric mutual information estimate between data distribution and cluster assignment. Although the optimization objective has a solid theoretical foundation it is hard to optimize. We propose an approximate optimization formulation that leads to an efficient algorithm with low runtime complexity. The algorithm has a single free parameter, the number of clusters to find. We demonstrate superior performance on several synthetic and real datasets.
Conference Title: DAGM: German Association For Pattern Recognition
Volume: 7476
Conference Dates: August 28-31, 2012
Conference Location: Graz, Austria
Publisher: Springer  
Location: Berlin, Heidelberg
Date Published: 2012-08-14
Start Page: 205
End Page: 215
Sponsor: Joanneum Research, Vexcel Imaging GmbH a Microsoft Company, MVTec Software GmbH, Siemens, SSI Schaefer.
DOI: 10.1007/978-3-642-32717-9_21
Open access: no
IST Austria Authors
  1. Christoph Lampert
    87 Lampert
Related IST Austria Work