Enabling Small-Data AI for Science

Developing AI (artificial intelligence) models for science often faces two critical challenges: training effective AI models given limited data and making robust competence-aware predictions under uncertainty. Scientific applications typically require modeling complex real-world systems or phenomena, for which the amount of available training data may be insufficient, and the cost for additional data acquisition may be formidably high. Consequently, effective strategies are needed for efficient data acquisition and experimental design as well as techniques for making robust and competence-aware predictions in the presence of substantial uncertainty. In this talk, we present recent advances in uncertainty quantification, optimal experimental design, and active learning, which altogether can enable small-data AI that can accelerate scientific discoveries across various fields. To demonstrate the advantages and potentials of these strategies, we will consider examples in systems biology, drug discovery, and material design.

Bio: 
Byung-Jun Yoon received the B.S. degree from the Seoul National University and the M.S. and Ph.D. degrees from the California Institute of Technology, all in Electrical Engineering. Since 2008, he has been with the Department of Electrical and Computer Engineering, Texas A&M University, where he is currently a professor. Yoon holds a joint appointment at Brookhaven National Laboratory, where he is a Scientist in Computational Science Initiative, Applied Mathematics Group. He received the NSF CAREER Award, the Best Paper Award at the 9th Asia Pacific Bioinformatics Conference and the 12th Annual MCBIOS Conference, and the SLATE Teaching Excellence Award from the Texas A&M University. Yoon’s main research interests lie in AI for Science, optimal experimental design, and objective-based uncertainty quantification. He is actively working on the development of these methods and their application to various scientific domains, including computational biology and materials science.
https://www.bnl.gov/staff/byoon
https://biomlsp.com/