Learning in Multi-Agent Systems

Telecommunications, economics, mobile robots, traffic simulation, electricity grids/smart grids ... are all examples of systems in which decentralisation of data and/or distribution of control is either required or inherently present. Moreover, in many such systems an exact model of the problem is not available, so exact planning methods are not applicable. The use of multiple interacting learning agents is a necessity to solve these problems. In such systems, where multiple agents are acting, the standard independent learning approaches don't scale up, so alternative approaches are required. Multi-agent Reinforcement learning is such a technique which is capable of dealing with multiple autonomous agents acting and learning in the same environment.