Research
I study machine learning methods that improve high-stakes decisions in scientific and industrial settings. My work combines representation learning, structured optimization, and systems design to build models that are both accurate and dependable.
Deep Learning for Pedestrian Safety
Developing trajectory prediction and surrogate safety measurement methods to evaluate pedestrian potential risk at non-signalized intersections.
Pedestrian Safety Trajectory Prediction Surrogate Safety
Spatiotemporal Transportation Data Modeling
Building spatial-temporal deep learning frameworks for large-scale road network speed prediction and mobility analytics under data sparsity.
Spatiotemporal Modeling Traffic Speed Prediction Variational Autoencoder
Generative AI and Mobility Safety
Applying deep learning and probabilistic modeling to transportation safety analysis, including crash risk estimation and motion heterogeneity modeling.
Crash Risk Estimation Probabilistic Modeling Transportation AI