Density Based Cluster Extension and Dominant Sets Clustering

With the pairwise data similarity matrix as input, dominant sets clustering has been shown to be a promising clustering approach with some nice properties. However, its clustering results are found to be influenced by the similarity parameter used in building the similarity matrix. While histogram equalization transformation of the similarity matrices removes this influence effectively, this transformation causes over-segmentation in the clustering results. In this paper we present a density based cluster extension algorithm to solve the over-segmentation problem. Specifically, we determine the density threshold based on the minimum possible density inside the dominant sets and then add new members into clusters if the density requirement is satisfied. Our algorithm is shown to perform better than the original dominant sets algorithm and also some other state-of-the-art clustering algorithms in data clustering and image segmentation experiments.

[1]  Longin Jan Latecki,et al.  Contour-based object detection as dominant set computation , 2012, Pattern Recognit..

[2]  Irfan A. Essa,et al.  A novel sequence representation for unsupervised analysis of human activities , 2009, Artif. Intell..

[3]  Aristides Gionis,et al.  Clustering aggregation , 2005, 21st International Conference on Data Engineering (ICDE'05).

[4]  Brendan J. Frey,et al.  Semi-Supervised Affinity Propagation with Instance-Level Constraints , 2009, AISTATS.

[5]  Dit-Yan Yeung,et al.  Robust path-based spectral clustering , 2008, Pattern Recognit..

[6]  Jitendra Malik,et al.  Spectral grouping using the Nystrom method , 2004, IEEE Transactions on Pattern Analysis and Machine Intelligence.

[7]  Huan Liu,et al.  Topic taxonomy adaptation for group profiling , 2008, TKDD.

[8]  D. Massart,et al.  Looking for natural patterns in data: Part 1. Density-based approach , 2001 .

[9]  Brendan J. Frey,et al.  Hierarchical Affinity Propagation , 2011, UAI.

[10]  Andrea Torsello,et al.  Beyond partitions: Allowing overlapping groups in pairwise clustering , 2008, 2008 19th International Conference on Pattern Recognition.

[11]  Brendan J. Frey,et al.  Non-metric affinity propagation for unsupervised image categorization , 2007, 2007 IEEE 11th International Conference on Computer Vision.

[12]  Ling Huang,et al.  Fast approximate spectral clustering , 2009, KDD.

[13]  Marcello Pelillo,et al.  A simple feature combination method based on dominant sets , 2013, Pattern Recognit..

[14]  Marcello Pelillo,et al.  Dominant Sets and Pairwise Clustering , 2007 .

[15]  Cor J. Veenman,et al.  A Maximum Variance Cluster Algorithm , 2002, IEEE Trans. Pattern Anal. Mach. Intell..

[16]  Andrea Torsello,et al.  Grouping with Asymmetric Affinities: A Game-Theoretic Perspective , 2006, 2006 IEEE Computer Society Conference on Computer Vision and Pattern Recognition (CVPR'06).

[17]  Jitendra Malik,et al.  Normalized cuts and image segmentation , 1997, Proceedings of IEEE Computer Society Conference on Computer Vision and Pattern Recognition.

[18]  Qi Xia,et al.  DSET++: A robust clustering algorithm , 2013, 2013 IEEE International Conference on Image Processing.

[19]  Delbert Dueck,et al.  Clustering by Passing Messages Between Data Points , 2007, Science.

[20]  Marcello Pelillo,et al.  Graph-based quadratic optimization: A fast evolutionary approach , 2011, Comput. Vis. Image Underst..

[21]  M. Pavan,et al.  A new graph-theoretic approach to clustering and segmentation , 2003, 2003 IEEE Computer Society Conference on Computer Vision and Pattern Recognition, 2003. Proceedings..

[22]  Michael I. Jordan,et al.  Dimensionality Reduction for Spectral Clustering , 2011, AISTATS.

[23]  Hans-Peter Kriegel,et al.  A Density-Based Algorithm for Discovering Clusters in Large Spatial Databases with Noise , 1996, KDD.