Christopher Qian, a doctoral student in the Department of Statistics at the University of Illinois Urbana-Champaign, has been named a recipient of the U.S. Department of Energy (DOE) Office of Science Graduate Student Research (SCGSR) Program award.
The SCGSR Program is a highly competitive award, with only 60 recipients selected from 48 universities nationwide for the 2023 Solicitation 1 cycle. The awardees will carry out part of their doctoral dissertation/thesis research in one of 13 DOE national laboratories, addressing scientific challenges central to the Office of Science’s mission areas across numerous research programs.
Qian’s research focuses on improving the probabilistic predictions of machine learning models. “Probabilistic predictions are useful because they help to capture the uncertainty present in the prediction, but it’s often the case that those predictions are overconfident, and underestimate the uncertainty,” he explained. “In my research proposal, I propose two potential approaches for developing novel methods to improve upon some of the currently used calibration methods in regression and classification.”
Upon learning of his selection, Qian expressed both surprise and joy. “I felt surprised that I was selected. They don’t give that many awards each year, and of the students selected historically, very few were in applied math or statistics,” he said. “However, I was quite happy to be selected, as I had put a lot of time in my application, and it gave me more confidence in my research proposal.”
Professor Feng Liang of the Department of Statistics and Qian’s advisor praised his work and its potential impact. “Chris’s thesis work is centered on Uncertainty Quantification (UQ), a critical aspect of scientific inquiry. In his research proposal, Chris delineated a series of post-processing calibration methods aimed at improving UQ without necessitating model retraining. In an era dominated by large-scale machine learning models, the issue of poor calibration poses a significant challenge to result trustworthiness. Chris’s contributions in this field have the potential to greatly enhance the trustworthiness of machine learning systems, with broad applications across various scientific domains,” she said.
The SCGSR award will allow Qian to collaborate with leading scientists and mentors at the Sandia National Laboratory. “Being able to work at Sandia is very helpful because of the support I can get from my mentor,” he said. “It’s a great opportunity to see what it’s like working at a national lab and learn about some of the national security research that goes on.”
Qian has been a beneficiary of the Sandia Academic Alliance Program since the inception of his thesis research, and this SCGSR award represents a collaborative effort between Illinois and Sandia.
The University of Illinois is a part of Sandia’s Academic Alliance Program, which partners with universities to solve significant national security challenges. This partnership allows Illinois faculty, researchers, and students to collaborate with researchers and leverage Sandia’s capabilities.
This opportunity aligns with the University of Illinois’ commitment to fostering impactful collaborations and nurturing the next generation of scientific innovators. Qian’s work contributes to advancements in machine learning and underscores the significance of collaborative partnerships in addressing critical challenges in scientific research.