Yuan-Sen Ting (OSU): AI+Science Schmidt Fellows Speaker Series

Organized by the University of Chicago’s Eric and Wendy Schmidt AI in Science Postdoctoral Fellowship Program.

Agenda
4:00pm – 4:45pm:  Presentation
4:45pm – 5:00pm:  Q&A
5:00pm – 5:30pm: Reception

Meeting location
William Eckhardt Research Center. Room 401
5640 S Ellis Avenue, Chicago, IL 60637
Map It

Abstract: The expansive, interdisciplinary nature of astronomy, combined with its open-access culture, makes it an ideal testing ground for exploring how Large Language Models (LLMs) can accelerate scientific discovery. In this talk, I will present our recent advances in applying LLMs to real-world astronomical challenges. Through self-play reinforcement learning, we demonstrate how LLM agents can conduct end-to-end research tasks in galaxy spectra fitting, encompassing data analysis, strategy refinement, and outlier detection—effectively mimicking human intuition and deep domain knowledge. Our agent, named Mephisto, rediscovered and analyzed the Little Red Dots, a new class of galaxies recently identified by the James Webb Space Telescope. While autonomous research agents like Mephisto could theoretically help analyze all observed sources, the cost of closed-source solutions remains prohibitive for large-scale applications involving billions of objects. To address this limitation, we are developing lightweight, open-source specialized models (AstroSage) and evaluating them against carefully curated astronomical benchmarks. Our research shows that specialized 8B-parameter LLMs can match GPT-4o’s performance on specific tasks when properly pretrained and fine-tuned. Despite ongoing challenges, we see potential in scaling up automated astronomical inference, which could transform how astronomical research is conducted.

Bio: Yuan-Sen Ting, Associate Professor of Astronomy at Ohio State University and an Adjunct Scientist at the Max Planck Institute for Astronomy. Previously, Yuan-Sen held positions as an Associate Professor of Astronomy and Computer Science at the Australian National University, and has held a Visiting Professor position at Tsinghua University and Universiti Malaya. Ting received his Ph.D. in Astrophysics and Astronomy from Harvard University. As a Malaysian native, he  had the privilege of living in seven countries and picking up six languages along the way, truly embracing the role of a global citizen. Ting  completed his concurrent Bachelor’s and Master’s degrees at the National University of Singapore and École Polytechnique in France on an Eiffel scholarship. During his studies he was honored with the Institute for Physics Medal and the National Academy of Science Award in Singapore, as well as a NASA Earth and Space Science Fellowship.

Post-Ph.D., Yuan-Sen was awarded a NASA Hubble Fellowship, Carnegie-Princeton Fellowship, and IAS Fellowship, enabling him to conduct postdoctoral research at Princeton’s Institute for Advanced Study. Ting  joined the Australian National University as a tenured faculty member before moving back to the U.S. to join OSU. He has also recognitions as the Humboldt Fellowship, the CCAPP Price Prize, and the ARC DECRA Fellowship. His research harnesses deep learning and artificial intelligence to advance data modeling and statistical inference in astrophysics. He specializes in extracting weak signals from massive datasets to tackle some of the field’s most challenging problems. This work has resulted in over 200 publications, garnering more than 8,000 citations, including a Nature cover story. As the leader of AstroMLab, a collaborative effort with Oak Ridge and Argonne National Laboratories, Ting guides a team of computer scientists and astrophysicists in developing specialized large language models and research agents for astrophysical applications.

Yuan-Sen is deeply committed to promoting science in his home country, Malaysia. He has organized two major astronomy conferences and a summer school there, writes columns for Sin Chew Jit Poh, a leading Chinese newspaper with over a million readers, and has given a TEDx talk in Malaysia and created two TED-Ed videos that have amassed over four million views combined.

Parking
Campus North Parking
5505 S Ellis Ave
Map It