Computational Intelligence and Multi-agent Games
Leader of the group Computational Intelligence and Multi-agent Games (SEN4): Han La Poutré.
The group works on the combination of two areas, consisting of computer science techniques and focus fields:
| 1. | Computational intelligence techniques, like evolutionary algorithms, adaptive algorithms, neural networks, graphical models and hybrid heuristics. |
| 2. | Competitive games in multiagent systems and optimization. |
Computational intelligence techniques
Fundamental research on computational intelligence techniques is also performed in the more conventional problem domains of optimization and classification. Here, the emphasis is on the mere development of these techniques, which in addition facilitates other research activities, like for multiagent systems.
Special attention is given to a novel type of neural networks, consisting of spiking neurons. Research aspects include learning rules, classification, binding, and credit assignment in neural networks via appropriate market mechanisms.
Research is also performed for improving and analyzing the performance of evolutionary algorithms and adaptive discrete algorithms, like for dynamic and online optimization problems, as well as novel types of optimization problems arising from multiagent systems applications.
Multi-agent Games: Learning Agents and Game Design
Interactions between self-interested agents in a multi-agent systems are often modelled by games, like negotiations, auctions, social dilemma games, and market games. Also, a prominent feature of an agent in such a system is the ability to learn. This combination forms a growing field of research within computer science, for both multi-agent systems (design of really-learning agents in games) and socio-economic sciences (simulation of learning agents).
In the SEN4 group, computational intelligence techniques are investigated, in order to build the internals of learning software agents participating in games (e.g. for e-business applications) as well as to simulate multi-agent systems governed by market games (socio-economic sciences).
Focus areas are among others the following:
- Intelligent algorithms and learning strategies for agents participating in competitive games (game theory), like in negotiations, auctions, social dilemma games (prisoner's dilemma) and dynamic pricing.
- Design and/or simulation of market games (market mechanisms, negotiations) in multiagent systems (for e-busines applications or socio-economic simulations).
Competitive games in multiagent systems are applicable in e.g. e-business, transportation logistics, health care planning, distributed recommendation, and economics.
Members
Hans Amman, Samuil Angelov, Sander Bohte, Peter Bosman, Anke Kristine Hutzschenreuter, Han Noot, Han La Poutré, Valentin Robu, Ivan Vermeulen, Mathijs de Weerdt, Mengxiao Wu.
Former members
List of former members
Events
SEN4 seminars and Journal Club meetings take place on a four-week basis, alternating every two weeks on Tuesdays.
This group is part of the cluster Software Engineering (SEN).
More information