My research has focused on elucidating the underlying nature of deep neural networks and, more importantly, applying it to real-world problems. In particular, I am interested in (1) “probabilistic neural networks” (e.g., Bayesian neural networks and ensemble methods); (2) “generalization” and “trustworthy machine learning” (e.g., uncertainty estimation and robustness); and (3) “exploratory analysis” (e.g., loss landscape visualization and Hessian eigenvalue spectrum).