Multi-agent reinforcement learning for coordination and problem structure

Project Details


Research in machine learning and reinforcement learning is increasingly moving towards multi-agent solutions, where distinct entities called agents together solve problems, such as routing in a network or load balancing on power grids. Learning in the presence of other dynamic, learning agents is challenging, and particularly difficult to scale to large groups of agents, without special measures. Current techniques that are scalable assume that agents only have weak interactions, and can solve the few conflicts encountered locally.

In settings where agents have stronger interactions, knowledge about these problem specific interactions can be used to structure coordination between the agents. This knowledge may be specified a priori, or learned during operation. However, current such algorithms require a global view of the system by each agent, limiting the decoupling between agents. Moreover, these approaches are only applicable in settings where all agents share the same interests.

In this project, we will investigate the automatic detection of interactions between agents, without requiring agents to have a full view of other agent's state and/or actions. Furthermore, we will validate the techniques developed from gained insights on settings that are fully cooperative, fully competitive and mixed.
Effective start/end date1/10/1230/09/16

Flemish discipline codes

  • Applied mathematics in specific fields


  • Databases
  • Evolution of language
  • Programming languages
  • Mobile Computing
  • Artificial Intelligence
  • Serious games
  • Web systems
  • Software agents