We offer The “Emergent Behavior In a Multi-Agent System with Reinforcement Learning” — EBIMAS study project, which is embedded as part of MicrocosmAI and is continuously offered at the IKW Osnabrück.
Here students work on projects about how artificial agents learn to communicate to reach a shared goal. What does an AI language look like? In this study project, we work together to develop and train embodied Reinforcement Learning agents in complex physical environments to examine emergent communication and behavior for cooperative tasks.
We implement our simulation via DeepMind’s MuJoCo engine and use TensorFlow for the RL algorithm. With some first milestones already reached, we are looking for motivated students to build on these results and help us in answering cutting-edge questions in an engaging work environment.
We are a perfect match if …
If you want to join the EBIMAS study project, you can sign up to the course on StudIP. There is also a separate 1-semester interdisciplinary seminar accompanying the study project. Since there are only a limited number of spots available, we may ask you to write a short letter of motivation to help us to better understand each candidate and determine the best fit for the program. Any further information can be found in the course details. We encourage you to apply and are looking forward to hearing from you!