Medlin, who expects to receive her Ph.D. in 2025, said she was excited to receive her first research grant.

Medlin will work with Krishnan Raghavan, an assistant computational mathematician in Argonne’s Mathematics and Computer Science (MCS) Division, to develop a machine learning algorithm that addresses imbalanced data challenges.

“Imbalanced data presents challenges to domain scientists across every domain; however, we have yet to find a common solution to the learning hurdles presented by this type of data. I’d like to change that,” Medlin said.

Specifically, under Raghavan’s mentorship and supported by the SCGSR program, Medlin and Raghavan will customize the ML algorithm to find rare fusion events amidst the vast sea of subatomic particle interactions detected by the ATLAS particle accelerator. The total number of events generated by ATLAS is in the order of millions while, in contrast, the number of fusion events is quite small.

“Conventional classification techniques used to extract rare fusion events involve manual and rudimentary tools requiring weeks or months of work and are marred by problems of imbalanced data,” Raghavan said.

Raghavan has been leading an investigation with researchers from the MCS and Physics Divisions at Argonne, using statistical and ML methods to identify patterns and classify the relevant events from reactions in an ionizing detector.

Medlin has been developing algorithms for classifying imbalanced data since summer 2022 when she first joined Argonne as an intern under Raghavan’s mentorship supported by the National Science Foundation’s Mathematical Science Graduate Research Internship program.

“I am deeply grateful to the SCGSR program for providing me the opportunity to work with Argonne National Laboratory experts on such an exciting problem while completing my dissertation research,” Medlin said.

Learn more about all eight awardees at Argonne, and find more information about the SCGSR program at the DOE website.