Shigenobu Kobayashi*, Masayuki Yamamura*, Isao Ono*, Keiki Takadama*, Jun Sakuma*,
Hiroshi Deguchi*, Hajime Kimura**, Kazuteru Miyazaki***, Sachiyo Arai****
(*Tokyo Institute of Technology, **Kyushu University, ***NIAD-UE, ****Kyoto University)
Mission
This team aims to establish methodological foundations for agent-based social systems
sciences from a viewpoint of adaptive systems. To attain such a goal, the following
themes are given.
Genetic Algorithms
- Developing diversity preserving strategies in generation alternation models
- Extending ƒ¿-domination strategy in multi-objective optimization
- Exploring optimization methods based on self-organization of state space
- Publishing a textbook, building an open software package, and implementing genetic algorithms on Soars
Reinforcement Learning
- Developing decision making strategies in distributed reinforcement learning
- Building a design theory for policy representation and rewards setting
- Integrating genetic algorithms and reinforcement learning
- Publishing a textbook, building an open software package, and implementing reinforcement learning algorithms on Soars
Collective Intelligence
- Analyzing case studies on ant colony systems, swarm systems, and so on
- Generalizing them as cooperating strategies in multi-agent systems
- Exploring distributed fault diagnosis in large scale systems
- Building an open software package and implementing collective intelligence algorithms on Soars
