POMDPs for dynamic troubleshooting

• Thesis title in Czech: POMDPs for dynamic troubleshooting
• Thesis title in English: POMDPs for dynamic troubleshooting
• Academic year of topic announcement: 2005/2006
• Thesis type: diploma thesis
• Thesis language: angličtina
• Department: Department of Theoretical Computer Science and Mathematical Logic (32-KTIML)
• Supervisor: Mgr. Marta Vomlelová, Ph.D.
• Author: hidden - assigned and confirmed by the Study Dept.
• Date of registration: 27.03.2006
• Date of assignment: 27.03.2006
• Date and time of defence: 02.02.2010 00:00
• Date of electronic submission: 02.02.2010
• Date of proceeded defence: 02.02.2010
• Opponents: RNDr. Jan Hric

Guidelines

Many stochastic processes can be modelled as decentralized POMDPs. Even
though a solution method is known, it is intractable in complex systems
(NEXP-complete).
The goal of this thesis is to review the state of the art in decentralized
POMDPs, to evaluate known heuristics for dynamic troubleshooting problems
and to try to design better heuristics.
As a starting point for testing, the model "Dinning philosophers with
falling sticks" [Ledl2004] will be used.

References

[Ledl2004] Svatopluk Lendl: Dining philosophers - Simulace víceprocesorového systému
s možností selhání a oprav, bakalářská práce, MFF, 2004

Brian Sallans: Reinforcement Learning for Factored Markov Decision
Processes, Ph.D. thesis, Department of Computer Science, University of
Toronto, 2002

R. Nair, M. Tambe, M. Yokoo, D. Pyndath, S. Marsella: Taming Decentralized
POMDPs: Towards Efficient Policy Computation for Multiagent Settings, In Proc. IJCAI, 2003.

I. Chades, B. Scherrer, F. Carpillet: A Heuristic Approach for Solving
Decentralized-POMDP: Assesment on the Pursuit Problem, In Proceedings of the Sixteenth ACM Symposium on Applied Computing, 2002.

Zilla Sinuany-Stern, Israel David, Sigal Biran: An Efficient Heuristic for a
Partially Observable Markov Decision Process of Machine Replacement, Computers & OR 24(2): 117-126 (1997)

Xavier Boyen, Daphne Koller: Tractable Inference for Complex Stochastic
Process, In Proc. UAI, pages 33--42, 1998