Course details

8.3579

Begleitseminar: Emergent Behavior in a Multi-Agent System with Reinforcement Learning

Interdisciplinary Course (Master)
SS 2023 Dr. Elia Bruni, Julius Mayer OFFLINE
4h/wk
6 ECTS
M.Sc modules:
CC-MP-IDC - Interdisciplinary Course
CS-MP-IDC - Interdisciplinary Course
Tue: 14-17
Wed: 14-17

This is an interdisciplinary seminar accompanying a study project. The group will work together to further develop a multi-agent system. The multi-agent system simulates a microcosm of simple artificial organisms (agents) in a puzzle-foraging scenario, trained with reinforcement learning. We make use of DeepMind’s MuJoCo physics engine to investigate emergent properties and dynamics between populations of agents, e.g. language, cooperation, and strategies. Note: This course is a 1-semester interdisciplinary seminar. Due to the high demand for last semester’s study project with the same topic, new students will be given the chance to join the current project group with the same topic for another two semesters. Prerequisites are competence in programming and either basic experience with simulation frameworks such as Unity or MuJoCo or basic knowledge of ML / RL. However, please note that due to the high demand, only a limited amount of new students can be admitted.