AI-MI junior investigator Poompol Buathong (Cornell University) presented “Evo-PU: Modeling Survivorship Bias with Positive-Unlabeled Learning for Protein Function Prediction” (with Zhongmou Chao, Ekaterina Selivanovitch, Susan Daniel and Peter Frazier) at the INFORMS Annual Meeting 2026 in San Francisco on November 2, 2026, in the invited session “Bayesian Optimization and Scientific Discovery.” AI-MI Associate Director Peter Frazier co-chaired the session with Giulia Pedrielli and also presents the talk “LLMs as Preference Oracles” at the meeting.
Source: INFORMS Annual Meeting 2026