Dr. Goran Radanovic

"Trustworthy Reinforcement Learning for Multi-agent Environments"

The deployment of Reinforcement Learning (RL) in complex, human-centered environments poses challenges that traditional RL frameworks do not account for. These challenges include performative feedback loops, corruption-prone feedback, and the need for accountability and justifiability. This talk provides an overview of our recent research aimed at addressing these challenges. I will present selected technical insights and results from our recent work, and conclude with a discussion on how these themes intersect with our ongoing research related to RL for large language models..

Bio: Goran Radanovic is a research group leader at the Max Planck Institute for Software Systems. In his research, he has studied various aspects of trustworthy AI, including robust and accountable sequential decision-making, incentive mechanism design, and fairness in machine learning. He has received several distinctions, including an Emmy Noether Fellowship. He earned his Ph.D. in Computer Science from EPFL in 2016. Prior to joining the Max Planck Institute for Software Systems, he was a postdoctoral researcher at Harvard University.

Termin als Datei downloaden und in den Kalender importieren.