Bergische Universität Wuppertal
Fakultät für Mathematik und Naturwissenschaften
Angewandte Informatik - Algorithmik


Support vector machine (SVM) learning


Tatjana Eitrich

Duration and funding

2004 - 2007


We investigate the optimization of parameters that are important for training support vector machines (SVM), in particular for highly unbalanced data sets.

Project-related publications

[1] Tatjana Eitrich and Bruno Lang. Efficient optimization of support vector machine learning parameters for unbalanced datasets. J. Comput. Appl. Math., 196(2):425--436, November 2006. [ Abstract ]
[2] Tatjana Eitrich and Bruno Lang. On the advantages of weighted L1-norm support vector learning for unbalanced binary classification problems. In Proc. IS'06, 3rd Intl. IEEE Conf. Intelligent Systems, September 4--6, 2006, University of Westminster, UK, pages 575--580. IEEE Computer Society, September 2006. [ Abstract ]
[3] Tatjana Eitrich, Bruno Lang, and Achim Streit. Customizing the APPSPACK software for parallel parameter tuning of a hybrid parallel support vector machine. In Giuseppe Di Fatta, Michael R. Berthold, and Srinivasan Parthasarathy, editors, Proc. PDM 2006, Workshop on Parallel Data Mining in conjunction with ECML/PKDD 2006, September 18--22, 2006, Berlin, Germany, pages 38--50, 2006. [ Abstract ]
[4] Tatjana Eitrich and Bruno Lang. Data mining with parallel support vector machines for classification. In Tatyana Yakhno and Erich J. Neuhold, editors, Proc. ADVIS 2006, 4th Intl. Conf. on Advances in Information Systems, October 18--20, 2006, Izmir, Turkey, volume 4243 of LNCS, pages 197--206, Berlin, 2006. Springer-Verlag. [ Abstract ]
[5] Tatjana Eitrich and Bruno Lang. On the efficient implementation of a serial and parallel decomposition algorithm for fast support vector machine training including a multi-parameter kernel. Int. J. Comput. Intell., 3(2):91--98, 2006. [ Abstract ]
[6] Tatjana Eitrich and Bruno Lang. On the optimal working set size in serial and parallel support vector machine learning with the decomposition algorithm. In Peter Christen, Paul J. Kennedy, Jiuyong Li, Simeon J. Simoff, and Graham J. Williams, editors, Data Mining and Analytics 2006, Proc. Fifth Australasian Data Mining Conference (AusDM2006), November 29--30, 2006, Sydney, Australia, volume 61 of Conferences in Research and Practice in Information Technology (CRPIT), pages 121--128. Australian Computer Society, Inc., 2006. [ Abstract ]
[7] Tatjana Eitrich and Bruno Lang. Parallel cost-sensitive support vector machine software for classification. In Ulrich H. E. Hansmann, Jan Meinke, Sandipan Mohanty, and Olav Zimmermann, editors, Proc. Workshop From Computational Biophysics to Systems Biology, June 06--09, 2006, Jülich, Germany, volume 34 of NIC Series, pages 141--144. John von Neumann Institute for Computing, Jülich, 2006. [ Abstract ]
[8] Tatjana Eitrich, Wolfgang Frings, and Bruno Lang. HyParSVM---a new hybrid parallel software for support vector machine learning on SMP clusters. In Wolfgang E. Nagel, Wolfgang W. Walter, and Wolfgang Lehner, editors, Parallel Processing, Proc. Euro-Par 2006, 12th European Conference on Parallel Computing, August 29--September 1, 2006, Dresden, Germany, volume 4128 of LNCS, pages 350--359. Springer-Verlag, 2006. [ Abstract ]
[9] Tatjana Eitrich and Bruno Lang. Analysis of support vector machine training costs for large and unbalanced data from pharmaceutical industry. In Proc. 1st ICGST Intl. Conf. Artificial Intelligence and Machine Learning (AIML-05), December 19--21, 2005, Cairo, Egypt, pages 58--64. ICGST, 2005. [ Abstract ]
[10] Tatjana Eitrich and Bruno Lang. Parallel tuning of support vector machine learning parameters for large and unbalanced data sets. In Michael R. Berthold, Robert Glen, Kai Diederichs, Oliver Kohlbacher, and Ingrid Fischer, editors, Computational Life Sciences: First International Symposium, CompLife 2005, Konstanz, Germany, September 25--27, Proceedings, volume 3695 of LNBI, pages 253--264, Heidelberg, 2005. Springer-Verlag. [ Abstract ]

Project-related theses

[1] Tatjana Eitrich. Dreistufig parallele Software zur Parameteroptimierung von Support-Vektor-Maschinen mit kostensensitiven Gütemaßen. Dissertation, Bergische Universität Wuppertal, Germany, January 2007.

See also

While performing this research, Tatjana Eitrich worked at the Zentralinstitut für Angewandte Mathematik, Forschungszentrum Jülich; see their home page.

University of Wuppertal
Faculty of Mathematics and Natural Sciences
Department of Mathematics and Computer Science
Applied Computer Science Group

Last modified:   Disclaimer