The first part focusses on learning through experience, of which reinforcement learning is the standard example. We start from simple single agent reinforcement learning (stateless RL) which we extent to deal with the interplay of multiple learning agents in the same environment. For this purpose and for the following part on evolutionary dynamics, basic concepts of Game Theory are introduced.
The second part provides an introduction to the principles of learning by imitation, modeled through evolutionary dynamics. It will explain what evolution is and how games can be used to model interactions between individuals in a population. It will show how these models can be used to study the evolution of cooperation in social dilemmas, the evolution of conventions like language or even the dynamics of cancer.
The course concludes with a project which can include, for those who are interested, experiments using the Khepera robots.
The aim of the course is to introduce the students to the field of learning in individual agents and learning in populations of agents and to prepare them for a Master thesis in these research areas.
He or she will learn the basic principles of both domains, the mathematical and computational methods and the typical problems they are trying to solve.
The students will also obtain a basic understanding of (evolutionary) game theory which will allow them to understand the standard literature in that field and the relevance of this domain to learning in general.
The students will obtain the skills to address independently problems within these fields.
In addition, they will be capable of presenting their work to an audience of specialists and non-specialists.
The course addresses two general areas of research : individual-based learning and social learning in populations.
The corresponding competences:
Knowledge and insight The student knows different Reinforcement Learning (RL) techniques. The student can judge different exploration strategies, is knowledgeable about the exploration/exploration trade off. The student has insight in the dynamics of multiagent reinforcement learning and the relevancy of the basic concepts of Game Theory in this context.
Application of knowledge and insight The student can solve a concrete problem using multiagent reinforcement learning technique.
Judgment ability The student can give arguments why a given problem is suited or not for RL.
Communication The students must complete 3 written assignments and 1 project in a team of 3 ot 4 students that needs to be presented.
Skills The student will be prepared to read autonomously the literature In this research domain.