Department of Civil & Urban Engineering at NYU Tandon Assistant Professor Eugene Vinitsky in #transportationengineering shared his expertise through a discussion on #multiagentsystems at the NYU Center for Data Science.
CDS-affiliated Assistant Professor Eugene Vinitsky from NYU Tandon shared insights on multi-agent reinforcement learning at a recent CILVR Seminar. His talk "PILAF: A stable foundation for empirically effective multi-agent reinforcement learning" challenged common assumptions about reinforcement learning algorithms in multi-player settings. Prof. Vinitsky demonstrated how single-agent reinforcement learning algorithms with appropriate regularization can outperform specialized multi-player algorithms. His research has practical applications in autonomous systems, including a simulated driver with below human-level crash rates. For the full schedule of upcoming CILVR seminars, see https://lnkd.in/ejw5tXyd #ReinforcementLearning #MultiAgentSystems #AIResearch