Section: Partnerships and Cooperations

International Initiatives

Inria Associate Teams

INDEMA

• Title: Intelligent Decision Making Mechanisms with Hidden Information, and Application to El ectricity Generation

• International Partner (Institution - Laboratory - Researcher):

  • NUTN (TAIWAN)

• Duration: 2012 - 2014

• See also: http://www.lri.fr/~teytaud/indema.html

• The objective of the project is three-fold:

  • Objective 1: Designing consistent iterative realistic algorithms for partially observable 1-player or 2-player games.

    $•$ Consistent algorithms (provably, asymptotically optimal in the computation time).

    $•$ Iterative a.k.a. anytime algorithms, improving its results as the computational time allowed increases and requiring little time to yield a decent answer. Most algorithms which survive decades are iterative.

    $•$ Realistic algorithms, i.e. suited to real-world settings.

  • Objective 2: Impressive visible applications, e.g. applications in games or puzzles, such as Minesweeper (on which we believe that much progress is still possible), Chinese Dark Chess, Kriegspiel, Phantom-Go, or card games. Games and puzzles offer nice frameworks to assess and make our research highly visible.

  • Objective 3: Big industrial applications. Having both mathematics and visible realizations in games and industrial applications might be considered as too ambitious. Yet, our strategy is to tackle e.g. the field of energy generation because: i) it is close from our past activities (thus reducing the warm-up time), yet with a new challenge, partial observability; ii) in real applications, many problems are simplified so that they boil down to fully observable problems, (e.g. through including tricks in the solvers); iii) our former achievements facilitate our contact with industry. Formally, we assume that mathematical analysis can be done on this (objective 1); that it will provide big results in games (objective 2) where many main programs are based on non-consistent algorithms; that these results will translate to real-world application.

    Our roadmap is:

    $•$ Check on simple versions of energy production problems whether the fully observable approximation holds true. We guess that in many cases it does not; the next point is to assess the loss of performance incurred;

    $•$ Experiment our algorithms on real industrial problems, considering both Taiwan-centered and Europe-Centered electricity generation problems in order to widen the scope of the analysis, enforcing the applicability of the approach.

Inria International Partners

Declared Inria International Partners

• Shinshu University (Professor Akimoto, Professor Tanaka, Professor Aguire). Partnership officialized via MOU signature between Inria and Shinshu University. Joint project funded by the Japanese governement.

• Dortmund University through the funded ANR project NumBBO.