[Computer-go] Fwd: Postdoc Vacancy on Computer Go and MCTS at Maastricht University

Erik van der Werf erikvanderwerf at gmail.com
Sat May 1 08:04:31 PDT 2010

---------- Forwarded message ----------
Postdoc Researcher on NWO Go4Nature project f/m
Maastricht, 6200 MD, The Netherlands, 38 hours per week
Universiteit Maastricht

Postdoc researcher on NWO Go4Nature project f/m

Job description

The Department of Knowledge Engineering (DKE), of the Faculty of
Humanities and Sciences (FH&S), has a vacancy for a postdoctoral
researcher (Postdoc) on the Go4Nature project funded by NWO.  The
project, supervised by Dr. Mark Winands,  researches the nature of
Monte-Carlo Tree Search and uses the increased insight to improve
search algorithms.

Two-person zero-sum games with perfect information such as Chess and
Checkers have been addressed by many Artificial Intelligence (AI)
researchers with great success, but the game of Go is a notable
exception. Despite significant attention the best computer Go programs
still lag far behind the level of professional players. Recently, a
new technique, called Monte-Carlo Tree Search (MCTS), started a
revolution and increased the level of play of computer Go programs
substantially. Whereas in first instance the focus of MCTS with
respect to Go has been on the practical application, the research
project addresses the problem of understanding the nature, the
underlying principles, of MCTS. A careful understanding of the nature
of MCTS will lead to (better) effective search algorithms. The two
interrelated research questions are: 1) How to formulate models that
increase the understanding how Monte-Carlo Tree Search (MCTS) works?
and 2) How to use the developed MCTS models to create effective search
algorithms? The research will be executed by the Postdoc (underlying
principles) together with a PhD researcher (application in Go).


Successful candidates are expected to meet the following requirements:
       -A Ph.D. degree in Computer Science, Artificial Intelligence,
Knowledge Engineering, or Mathematics. Candidates expecting to
graduate by 1 September 2010 can also apply.
       - Extensive knowledge in Intelligent Search Techniques and/or
Games &   AI
       - A creative and dynamic personality.
       - An enthusiastic and cooperative working attitude.
       - Good communicative skills.
       - Good didactic abilities.


The DKE is responsible for the Bachelor programme Knowledge
Engineering of Maastricht University. Moreover, the DKE is responsible
for the Master programmes Artificial Intelligence and Operations
Research of the School for Information Technology (SIT) of the
transnational University Limburg (tUL), a collaboration between
Maastricht University and Hasselt University. The research in the
Department of Knowledge Engineering (DKE) is organized in three
research groups: “BioMathematics and BioInformatics” (BMI), “Robots,
Agents and Interaction” (RAI) and “Networks and Strategic
Optimization” (NSO). The candidate will be employed in the NSO
research group. For the NSO research group the emphasis is on
Knowledge Engineering topics related to the study of competition. The
group has a solid background in two very different approaches to
competition. While part of the group has a lot of expertise in
examining competition in board games, say “game practical” modelling,
the other part is specifically experienced in game theoretical

Conditions of employment

Assignment of the researcher to a salary scale will depend on
knowledge and experience (€ 2379 - € 3755);
Employment basis: 3 years
Maximum hours per week: 38

Additional Information

Additional information about the vacancy can be obtained from:
Dr. M.H.M. Winands
Tel.: +31 43 38 82728
Email: m.winands at maastrichtuniversity.nl


You can apply for this job before 15-06-2010 by sending your application to:

Maastricht University
Dr. M.H.M. Winands
P.O. Box 616
6200 MD Maastricht
The Netherlands
Email: m.winands at maastrichtuniversity.nl

When applying for this job always mention the vacancy number AT2010.88

More information about the Computer-go mailing list