Some Software Packages
(from Shi Zhong's research)

All programs listed below are distributed under GNU GPL license, thus are free for you to use or distribute. But please be aware that these programs come with absolutely no warranty what so ever, and in no way will I be held responsible for loss of properties due to the use of these software packages. So use it at your own risk.

Model-based Text Clustering (in Matlab)

• Brief description
  • The package contains model-based hard k-means and soft EM clustering algorithms for text applications. Probabilistic models implemented include multivariate Bernoulli, multinomial, and von Mises-Fisher models. Deterministic annealing version of all the above algorithms are also included.
• Source code
  • [textclust.zip] For usage, read the README file.
• Text data in Matlab format
  • [docdata.zip] For a description of this data, see the first reference below.
• References
  • Shi Zhong and Joydeep Ghosh, "Generative model-based clustering of documents: a comparative study," Knowledge and Information Systems (KAIS), Vol. 8, 2005. pp. 374-384.
    • An earlier workshop version: Shi Zhong and Joydeep Ghosh. A comparative study of generative models for document clustering. In SDM Workshop on Clustering High Dimensional Data and Its Applicatons, San Francisco, CA. May 2003.
  • Shi Zhong and Joydeep Ghosh. A Unified Framework for Model-based Clustering. Journal of Machine Learning Research (JMLR). vol. 4, pp. 1001-1037. November, 2003.
  • Shi Zhong and Joydeep Ghosh. Scalable, balanced model-based clustering. In SIAM Int. Conf. Data Mining (SDM 2003), pp. 71-82, San Francisco, CA. May 2003.

Online Spherical K-Means Clustering (OSKM, in C & Matlab)

• Source code
  • [C code | Matlab interface | Utility program that can convert text data from Matlab format to CCS format required by the C code.]
• References
  • Shi Zhong, "Efficient Online Spherical K-means Clustering," In Proc. IEEE Int. Joint Conf. Neural Networks (IJCNN 2005), Montreal, Canada, July 31-August 4, 2005. pp. 3180-3185.

Coupled Hidden Markov Models (CHMM, in Matlab)

• Source code
  • [dchmm.zip]
• References
  • Shi Zhong and Joydeep Ghosh. Coupled Hidden Markov Models. Tech. Report, ECE Dept., University of Texas at Austin. Austin, Texas, USA. June, 2001.
  • Shi Zhong and Joydeep Ghosh. HMMs and coupled HMMs for multi-channel EEG classification. In Proc. IEEE Int. Joint Conf. on Neural Networks (IJCNN'02), vol. 2, pp. 1154-1159, Honolulu, Hawaii. May 2002.

©Shi Zhong, 2003-2005