AI Research Talks @ Percepta #2
Welcome back to Research Talks at Percepta, where leading AI researchers share their work with the NYC community of researchers, engineers, and operators.
For our next session, we're excited to host our very own Eugene Vinitsky presenting a newly published Nature paper he coauthored: Scalable decision-making for games of imperfect information, which introduces the first superhuman AI for Stratego.
Most real-world decisions involve hidden information, which breaks the reinforcement learning and search methods that mastered chess and Go. Even multimillion-dollar industrial research efforts had failed to reach top-human-level play at Stratego, a board wargame with hidden information on a massive scale. Eugene and his collaborators at CMU, MIT, and Stanford (Samuel Sokota, Hengyuan Hu, Zhiyuan Fan, Zico Kolter, and Gabriele Farina) built Ataraxos, which beat the strongest Stratego player in the world and did it on a budget that was orders of magnitude smaller.
Full paper can be found here: https://www.nature.com/articles/s41586-026-11036-y
About Eugene
Eugene is a Research Scientist at Percepta and an Assistant Professor of Civil and Urban Engineering at NYU Tandon, with an affiliation in Computer Science. He received his PhD from UC Berkeley, and then worked as a research scientist in the Apple Special Projects Group. At Berkeley, he built the reinforcement learning algorithms and simulators used to deploy autonomous cruise controllers on real highways, and he has also been a visiting researcher at Facebook AI and interned at DeepMind. His research focuses on how complex, human-like behavior emerges from interaction between groups of learning agents.
Who it's for
Researchers, engineers, and operators working in or interested in AI.
Agenda
Light bites, drinks, and networking
Eugene's talk
Q&A with the audience
📅 Thursday Oct 22
📍 Address shared with approved guests
Space is limited, so registration is by approval. Request to join and we'll confirm your spot by email.