Dirk Ormoneit

Address:

Department of Computer Science
Gates Building, 1A-148
Stanford University
Stanford, CA 94305-9010

phone: (650) 725-8797
fax: (650) 725-1449

e-mail: ormoneit@robotics.stanford.edu

New papers on reinforcement learning, particle filters, and human motion:

• D. Ormoneit and P. W. Glynn. Kernel-based reinforcement learning in average-cost problems, 2001. Submitted for publication.
• P. W. Glynn and D. Ormoneit. Hoeffding's Inequality for Uniformly Ergodic Markov Chains.
  In Statistics and Probability Letters, 2001. To appear.
• D. Ormoneit, C. Lemieux, and David J. Fleet. Lattice Particle Filters, 2001. In Uncertainty in Artificial Intelligence, 2001. To appear.
• D. Ormoneit, T. Hastie, and M.Black. Functional analysis of human motion data, 2000. Submitted for publication.

Currently, I am research associate at the Department of Computer Science at Stanford University, working in the DAGS group of Daphne Koller. Previously, I was postdoctoral student at the Department of Statistics at Stanford, where I am working with Trevor Hastie, and doctoral student at the chair for ``Theoretical Computer Science and Foundations of Artificial Intelligence'' at the Technische Universität München, where I received my doctoral (Dr. rer. nat.) and graduate degrees (Dipl.-Inform.) in the years 1998 and 1994, respectively. From 1994 to 1995, I held a position as a strategical analyst in the Corporate Planning Department of Siemens AG, Munich. During 1996 I was a visiting scholar the Department of Economics at UCSD.

The focus of my research is on machine learning. Specific areas of interest include reinforcement learning, particle filters & computer vision, data mining, financial forecasting, neural networks, and dynamic Markov models. Additional areas of interest include electronic commerce and mathematical finance with emphasis on option pricing.

I am co-organizing the Nobots seminar on Machine Learning.

More about my research

• Reinforcement Learning & Control (with Peter Glynn, Carlos Guestrin, and Saunak Sen; also try the new DAGS MDP page ),
• Particle Filters & Computer Vision (with Christiane Lemieux, David Fleet Xerox PARC, Michael Black, and Hedvig Sidenbladh)
• Data Mining (with Trevor Hastie and Volker Tresp)
• Computational Finance (with Halbert White and Ralph Neuneier)

Papers and Publications (see also the BIBTEX file)

Books

• D. Ormoneit. Probability Estimating Neural Networks, Shaker Verlag, Doctoral Thesis, Institut für Informatik, Technische Universität München, 1998. ISBN 3-8265-3723-8.
  You can find a summary here.

Book Chapters

• D. Ormoneit and R. Neuneier. Conditional value at risk. In Computational Finance. MIT Press, 1999, pages 41-52.
  An older version of this paper is available online.

Journal Articles

• P. W. Glynn and D. Ormoneit. Hoeffding's Inequality for Uniformly Ergodic Markov Chains. Statistics and Probability Letters, 2001. To appear.
• D. Ormoneit and S. Sen. Kernel-based reinforcement learning. Machine Learning, 2002. To appear.
  An older technical report is available here.
• D. Ormoneit. A regularization approach to continuous learning with an application to financial derivatives pricing. Neural Networks, 12(10):1405-1412, 1999.
  An older version of this work as a working paper is available here.
• D. Ormoneit and H. White. An efficient algorithm to compute maximum entropy densities. Econometric Reviews, 18(2):127-140, 1999.
  The table of lambda- and mu-values which is mentioned in this paper is here and the C++ program for the numerical integration is here.
• D. Ormoneit and V. Tresp. Averaging, maximum penalized likelihood and Bayesian estimation for improving Gaussian mixture probability density estimates. IEEE Transactions on Neural Networks, 9(4):639-650, 1998.

Reviewed Conference Articles

• C. Lemieux, D. Ormoneit, and David J. Fleet. Lattice Particle Filters. In Uncertainty in Artificial Intelligence, 2001. To appear.
• C. Guestrin and D. Ormoneit. Robust Combination of Local Controllers.. In Uncertainty in Artificial Intelligence, 2001. To appear.
• U. Chajewska, D. Koller, and D. Ormoneit. Learning about an Agent's Utility Function by Observing Behavior.. In International Conference on Machine Learning, 2001. To appear.
• D. Ormoneit and P. Glynn. Kernel-based reinforcement learning in average-cost problems: An application to optimal portfolio choice. In Advances in Neural Information Processing Systems 13. The MIT Press, 2001.
• D. Ormoneit, H. Sidenbladh, M. J. Black, and T. Hastie. Learning and tracking cyclic human motion. In Advances in Neural Information Processing Systems 13. The MIT Press, 2001.
• D. Ormoneit, H. Sidenbladh, M. J. Black, T. Hastie, and D. Fleet. Learning and tracking human motion using functional analysis. In IEEE Workshop on Human Modeling, Analysis and Synthesis, 2000.
• D. Ormoneit and T. Hastie. Optimal kernel shapes for local linear regression. In S. A. Solla, T. K. Leen, and K-R. Müller, editors, Advances in Neural Information Processing Systems 12, pages 540-546. The MIT Press, 2000.
• D. Ormoneit and R. Neuneier. Experiments in predicting the German stock index DAX with density estimating neural networks. In Proceedings of the IEEE/IAFE 1996 Conference on Computational Intelligence in Financial Engineering (CIFEr), pages 66-71, 1996.
• D. Ormoneit and V. Tresp. Improved Gaussian Mixture Density Estimates Using Bayesian Penalty Terms and Network Averaging. In D. S. Touretzky, M. C. Mozer, and M. E. Hasselmo, editors, Advances in Neural Information Processing Systems 8, pages 542-548. The MIT Press, 1996.
• D. Ormoneit and R. Neuneier. Reliable Neural Network Predictions in the Presence of Outliers and Non-Constant Variances. In A. N. Refenes, Y. Abu-Mustafa, J. Moody, and A. Weigend, editors, Neural Networks in Financial Engineering. World Scientific Publis hing, pages 578-587, 1995.
• R. Neuneier, F. Hergert, W. Finnoff, and D. Ormoneit. Estimation of Conditional Densities: A Comparison of Neural Network Approaches Proceedings of the International Conference on Artificial Neural Networks, 1:689-692, 1994.

Technical Reports & Working Papers

• P. W. Glynn and D. Ormoneit. Hoeffding's Inequality for Uniformly Ergodic Markov Chains, 2001. Submitted for publication.
• D. Ormoneit and P. W. Glynn. Kernel-based reinforcement learning in average-cost problems, 2001.
• C. Lemieux, D. Ormoneit, and David J. Fleet. Lattice Particle Filters, 2001. Submitted for publication.
• D. Ormoneit, T. Hastie, and M.Black. Functional analysis of human motion data, 2000. Submitted for publication.
• D. Ormoneit and T. Hastie. Optimal Kernel Shapes for Local Linear Regression Technical Report 1999-11, Department of Statistics, Stanford University, 1999. Submitted for publication.
• D. Ormoneit and S. Sen. Kernel-Based Reinforcement Learning Technical Report 1999-8, Department of Statistics, Stanford University, May 1999.
• Forschungsbericht Künstliche Intelligenz, Technische Universität München 1997, FKI-220-97: An Efficient Algorithm to Compute Maximum Entropy Densities
  The table of lambda- and mu-values which is mentioned in this paper is here and the C++ program for the numerical integration is here.
• D. Ormoneit. Arbeits- und Ergebnisbericht zum ersten Fortsezungsantrag auf Weiterführung des Graduiertenkollegs, Institut für Informatik, Technische Universität München 1997, Kapitel 1.3.7: Computational Economies, Wahrscheinlichkeitsdichten und Neuronale Netze zur Entscheidungsmodellierung in Multiagentensystemen
• D. Ormoneit and V. Tresp Forschungsbericht Künstliche Intelligenz, Technische Universität München 1997, FKI-205-95: Improved Gaussian Mixture Density Estimates Using Bayesian Penalty Terms and Network Averaging .

Review Activities

• Journal of Machine Learning, 1999-2000.
• IEEE Transactions on Neural Networks, 2000.
• Annals of Statistics, 1998-2000.
• Computer Vision and Pattern Recognition, 2000.
• Journal of Forecasting, 1996.
• ICONIP, 1996.

Some Talks

• Yale University, Department of Computer Science, 2001.
• Columbia University, Department of Computer Science, 2001.
• Siemens Research Center, Munich, 2000.
• IEEE Workshop on Human Modeling, Analysis and Synthesis. South Carolina, 2000.
• Department of Computer Science, Stanford University, 2000.
• Xerox Palo Alto Research Center, 2000.
• IBM Almaden Research Center, 2000.
• IMS/Beroulli Conference, Guanajuato, Mexico, 2000.
• Machine Learning Workshop, Snowbird, 2000.
• Computational Learning Laboratory, Stanford University, 1999.
• Siemens Research Center, Munich, 1999.
• Department of Statistics, Stanford University, 1999.
• Continuous Learning Workshop, Neural Information Processing Systems, Breckenridge, 1998.
• Department of Finance, University of Augsburg, 1998.
• Department of Statistics, Stanford University, 1997.
• Machine Learning Workshop, Snowbird, 1997.
• Department of Finance, MIT Sloan School of Business, 1997.
• Neural Networks in the Capital Markets, London, 1995.
• International Conference on Artificial Neural Networks, Sorrento/Italy, 1994.
• Department of Economics, University of California, San Diego, 1993.

• About this document ...