Research spans tensor-based learning, latent-variable graphical models, and high-dimensional inference. Develops identifiability theory, structure discovery, and convex estimators; leverages external networks and transfer learning to improve variable selection, prediction, and interpretability.