CALL FOR PAPERS

               Machine Learning for Sensor Planning
               http://www.cs.york.ac.uk/~grzes/mlsp/

               An ICML 2007 Workshop and Competition
               Corvallis, Oregon, USA, June 24, 2007

MOTIVATION

Sensor planning is the source of very complex decision problems,
as a result of partial observability and the need to reason about
the information gain from sensing actions. Also, sensor planning
is a fundamental enabling technology which has broad applicability
to future intelligent systems in transportation, emergency services
and security.

The challenges posed by the domain of sensor planning are highly
relevant for current machine learning research. ML techniques that
can be applied to sensor planning are wide-ranging, and include
reinforcement learning, evolutionary approaches, model learning
and dynamic programming, etc.

The MLSP workshop aims to encourage multi-disciplinary research
to address these important and challenging problems. It will bring
together researchers from a wide range of ML-related areas. In
addition, we provide a sensor planning benchmark platform to enable
the participants and the wider ML community to compare the
performance of the various techniques.

THE SENSOR PLANNING COMPETITION

In order to encourage a practical focus of workshop contributions,
and to provide a benchmark for direct comparison of algorithm
performance, we have defined and implemented a sensor planning
challenge scenario. Researchers are encouraged to submit programs
for the competition, and a competition will take place at the
workshop. Cash prizes totalling 00 will be awarded to winning
programs.

To participate at the workshop, software submission is not necessary.
It is possible to submit a paper without software, and to submit
software without a paper.

The Benchmark Challenge Scenario

There has been an earthquake over a large urban region. The search for
survivors focuses on a heavily populated