W375 Westgate Building
10:00am
Biological systems are high-dimensional, structured, and inherently dynamic, posing fundamental challenges for AI models in data-limited scientific settings. My research develops structured and scalable AI methods for learning such systems across molecular and cellular scales, with an emphasis on reliability, identifiability, and computational efficiency. I begin with advances in L0-based sparse modeling for high-dimensional structure recovery, improving model selection accuracy under small-sample and high-correlation regimes. At the cellular scale, these methods uncover transcription factor collaborations and strengthen gene regulatory network inference. To capture regulatory interactions and temporal dynamics, we develop scalable linear ODE models that enable dynamic system analysis through GPU-accelerated computation. These models move beyond static association toward system-level simulation and in silico perturbation, forming the basis for virtual cellular modeling. At the molecular scale, I lead an NIH-funded program developing AI methods for large-scale virtual screening and blood-brain barrier permeability prediction
Additional Information:
Yijie Wang is an Associate Professor in the Department of Computer Science at Indiana University Bloomington, where he has been on the faculty since 2019. His research develops foundational AI and machine learning methods for modeling complex, high-dimensional biological systems. He focuses on interpretable, structured learning through sparsity, advancing algorithmic approaches to uncover gene regulatory mechanisms. His work bridges AI, applied mathematics, and biomedical sciences, with translational applications in AI-driven drug discovery. He is the recipient of the NIH MIRA (R35) Award in 2022 and an NIA R01 grant in 2026.
Details...
