Paul Wagner

I work as a post-graduate researcher in the Computational Cognitive Systems group. My main research interest is to develop and apply reinforcement learning (approximate dynamic programming) and related optimal control methodology in the context of simulated robotics, virtual actors, and programming by demonstration.

Publications

Reinforcement learning

Wagner, P.

A reinterpretation of the policy oscillation phenomenon in approximate policy iteration
To appear in Advances in Neural Information Processing Systems (NIPS).
Download: nips version [pdf], extended version [pdf], poster [pdf]

Wagner, P. (2010)

On the stability of reinforcement learning under partial observability and generalizing representations
Master's thesis, Aalto University School of Science and Technology, Department of Information and Computer Science, Espoo, Finland.

Independent variable group analysis

See the project homepage for details.

Alhoniemi, E., Honkela, A., Lagus, K., Seppä, S. J., Wagner, P., & Valpola, H. (2007)

Compact modeling of data using independent variable group analysis
IEEE Transactions on Neural Networks, 18(6), 1762-1776.

Lagus, K. Alhoniemi, E., Seppä, J., Honkela, A., & Wagner, P. (2005)

Independent Variable Group Analysis in Learning Compact Representations for Data
In Proceedings of the International and Interdisciplinary Conference on Adaptive Knowledge Representation and Reasoning (AKRR'05), Espoo, 15.-17. June 2005. Pages 49-56.

Contact information

Visiting address	Room A335 Computer Science Building Konemiehentie 2, Espoo, Finland
Mailing address	Aalto University School of Science and Technology Department of Information and Computer Science P.O. Box 15400, FI-00076 Aalto, Finland
E-mail	pwagner at cis dot hut dot fi