Cluster - Fuzzy and Probabilistic Clustering

Download

Note: The table package contains some auxiliary programs for preprocessing the data files.

Description

Cluster is a set of programs to do probabilistic clustering (expectation maximization algorithm to find a mixture of Gaussians) and fuzzy clustering (fuzzy c-means algorithm, Gustafson-Kessel algorithm, and Gath-Geva / FMLE algorithm) and to execute the induced set of clusters on new data. The programs are highly parameterizable, so that a large variety of clustering approaches can be carried out. Since version 2.4 the program also contains a competitive learning / learning vector quantization mode, which will eventually supersede the learning vector quantization programs. A brief description of how to apply these programs can be found in the file cluster/ex/readme in the source package.

If you have trouble executing the programs on Microsoft Windows, check whether you have the Microsoft Visual C++ Redistributable for Visual Studio 2022 (see under "Other Tools and Frameworks") installed, as the programs were compiled with Microsoft Visual Studio 2022.

An extensive treatment of the clustering approaches supported by the programs can be found in my habilitation thesis:

• Prototype-based Classification and Clustering
  Christian Borgelt
  Habilitation Thesis, University of Magdeburg, Germany 2005/2006

A description of the ideas underlying the neural network based update methods can also be found in:

• Speeding Up Fuzzy Clustering with Neural Network Techniques
  Christian Borgelt and Rudolf Kruse
  Proc. 12th IEEE International Conference on Fuzzy Systems (FUZZ-IEEE'03, St. Louis, MO, USA)
  IEEE Press, Piscataway, NJ, USA 2003
  fieee_03.pdf (172 kb) fieee_03.ps.gz (75 kb) (5 pages)

A description of the regularization methods can also be found in:

• Shape and Size Regularization in Expectation Maximization and Fuzzy Clustering.
  Christian Borgelt and Rudolf Kruse.
  Proc. 8th European Conf. on Principles and Practice of Knowledge Discovery in Databases (PKDD 2004, Pisa, Italy), 52-62.
  Springer, Heidelberg, Germany 2004
  pkdd_04.pdf (235 kb) pkdd_04.ps.gz (321 kb) (11 pages)

A description of learning vector quantization with size and shape parameters can be found in:

• Fuzzy Learning Vector Quantization with Size and Shape Parameters
  Christian Borgelt, Andreas Nürnberger, and Rudolf Kruse
  Proc. 15th IEEE International Conference on Fuzzy Systems (FUZZ-IEEE'05, Reno, NV), CD-ROM
  IEEE Press, Piscataway, NJ, USA 2005
  fieee_05.pdf (184 kb) fieee_05.ps.gz (126 kb) (6 pages)