|
Click on a theme or a project in the table below for more information.
Project leader:
Prof.dr. Krzysztof Apt (CWI)
Consortium:
CWI
Total FTE: 3.0 (heads: PD 3)
Key BRICKS publications:
| • |
M.M. de Weerdt, Y. Zhang and T. Klos. Distributed task allocation in social networks. In AAMAS 2007, pp. 488-495, 2007.
|
| • |
T. Klos, G.J. van Ahee. Evolutionary dynamics for designing multi-period auctions. In AAMAS 2008, pp. 1589-1592, 2008.
|
| • |
K.R. Apt, F. Arbab and H. Ma. A Distributed Platform for Mechanism Design. In. Proc. of the International Conference on Computational Intelligence for Modelling, Control and Automation, IEEE Society, pp. 1-6, 2008.
|
| • |
K.R. Apt, V. Conitzer, M. Guo and E. Markakis. Welfare Undominated Groves Mechanisms. In WINE 2008, pp. 426-437, 2008.
|
| • |
V. Robu, H. Noot, H. La Poutré, W.-J. van Schijndel. An interactive platform for auction-based allocation of loads in transportation logistics. In AAMAS 2008, pages 3-10, 2008.
|
|
Project IS7: Distributed Implementations of Adaptive Collective Decision Making
One of the main themes in economics has been the study of markets
and of the rational agents participating in them. This has led to
a development of several fundamental methods and techniques
concerned with collective decision making, notably mechanism
design. Two important developments in computer science gave a new
impetus and led to a renewed interest in these economic theories.
One of them is grid computing. The other is electronic commerce in
which one needs to deal with a whole range of issues including
security, trust, negotiation protocols and strategies, reliable
payment mechanisms, business-to-business and business-to-consumer
transactions, etc. In addition, the characteristics of many of
these issues point to the need to study repeated mechanisms, and
the adaptive strategies agents may employ to be successful in
sequences of interactions.
Distributed implementations
The aim of this subproject is to develop generic, distributed and
fault-tolerant implementations of various forms of mechanism
design, including Vickrey-Clarke-Groves (VCG) mechanism. Such
implementations should be robust against sophisticated forms of
manipulation available in a distributed setting, such as
submission of multiple bids from multiple email addresses, or, in
the case of rings of hierarchical networks, falsification of
information about other agents that needs to be passed further. In
addition such an implementation should provide support for various
forms of communication and network topology.
Sequential mechanism design
This subproject aims at a study of mechanism design in the case
when the agents submit their reports not simultaneously but
sequentially. In this setting other dominant strategies than
truth-telling may exist. It will be explored how this can be used
to minimize in some situations the overall collected tax.
From the point of view of game theory it will be investigated how
players in addition to maximizing their own payoff can also aim at
maximizing payoff of (some) other players. This additional
objective is natural in the games which the players may or will
play repeatedly.
Adaptive strategies
This subproject concentrates on adaptive strategies for repeated
interactions. Such strategies take into account the fact that for
boundedly rational agents, knowledge of future games may only
become available over time, and learning is based on experiences
in past games. We use Computational Intelligence techniques for
designing such adaptive strategies.
Industrial cooperation
In the context of BRICKS we maintain close contacts with a wide
variety of internationally well-renowned research institutes,
including University of Amsterdam, Leiden University, TU Delft, TU
Eindhoven, Chinese University of Hong-Kong, École Polytechnique,
Lausanne (Switzerland), University of Padova (Italy), University of
Bielefeld (Germany), Southampton University (UK), SICS (Swedish
Institute for Computer Science), CMU (USA).
Highlights
As this project is part of the third phase of the BRICKS program
(financed through the second open round July 2006) and started
October 2006, challenges rather than results are presented.
Likewise, no BRICKS key publications are currently available.
Research challenges
- Realize a generic, distributed implementation of various types of mechanism design that provides support for different forms of communication and network topology, is reliable w.r.t. sophisticated forms of manipulation, is fault-tolerant.
- Identify and implement strategies that aim at minimization of collected taxes both in the simultaneous and sequential mechanism design.
- Identify and implement adaptive strategies for repeated mechanisms.
IS7 Researchers funded by BRICKS
- Dr. Arantza Estévez-Fernández (CWI)
- Dr. Huiye Ma (CWI)
- Dr. Tomas Klos (CWI)
- Dr. Vangelis Markakis (CWI)
- Drs. Valentin Robu (CWI)
- Dr. Dominik Wojtczak (CWI)
Other researchers involved
- Prof.dr. Krzysztof Apt (CWI)
- Prof.dr. Farhad Arbab (CWI)
- Prof.dr. Han La Poutré (CWI)
For more information, please refer to the publications and posters of this project.
|
