Events

Reinforcement Learning (RL) has emerged as a fundamental tool for the design of artificial agents possessing optimality and safety guarantees. A series of results has established that the behaviour learnt by RL agents interacting with one another can be understood using the tools of evolutionary game theory – for example, one can investigate conditions under which agents’ learnt policies converge to Evolutionarily Stable Strategies of the underlying game. Issues such as the emergence of cooperation, the tragedy of the commons, partner selection, population structure, and the exploration/exploitation trade-off are key questions in the study of both artificial intelligence and evolutionary biology and can be described using similar underlying mathematical frameworks. The aim of this workshop is to explore recent developments at this interface, to discuss open questions, and to bring together researchers from both communities.