Yanshen Sun began her academic journey with a bachelor’s degree in geographic information science from Zhejiang University, China. She went on to earn two master’s degrees, in geography and computer science, at Virginia Tech. 

Sun was drawn to the Sanghani Center as a Ph.D. student in the Department of Computer Science while attending a welcoming event where a presentation highlighted the center as a community committed to high-quality, interdisciplinary research. 

“I immediately knew I wanted to be part of that conversation. What I love most about being a Sanghani Center student is the opportunity to engage with diverse thinkers and researchers -- every discussion pushes my ideas further and opens up new avenues for exploration,” said Sun, who is advised by Chang-Tien Lu.

Sun’s research focuses on dynamic graph neural networks (GNNs) applied to spatiotemporal data, such as traffic networks and brain networks. 

One example is performing epilepsy diagnosis by analyzing electroencephalograms (EEGs). EEG data, as brain signals measured outside human scalp are usually noisy and make it hard to generalize models trained with some people’s data to other people’s data. 

“To overcome these challenges, we designed a self-supervised training pipeline to denoise the EEG signals and remove their differences of domains across different subjects. We then apply a score-based spatiotemporal transformer to determine whether specific EEG segments indicate epilepsy,” said Sun.

“I started with traffic data since it perfectly marries my background in GIS with my passion for urban computing,” she said. “As I attended seminars and chatted with peers, I became equally fascinated by the potential of these methods in medical and biological informatics.”

Sun has published a number of papers, collaborating with her supervisor Lu, among others. The most recent is “Downscaling Precipitation with Bias-informed Conditional Diffusion Model,” in proceedings of the 2024 IEEE International Conference on Big Data (IEEE BigData), in Washington, D.C., December 2024.

Earlier that month, she received the Best Student Paper Award at the IEEE International Conference on Bioinformatics and Biomedicine (BIBM) for “Empowering Cross-Patient Epilepsy Diagnosis from Diverse-Sampling Low-Quality EEG Signals."

Included among her other published works are:

Sun is projected to graduate in May 2025 and plans to begin her career as a research scientist at Meta, a return offer from an internship she did there.