Electrical Engineering Department
Research Interests: Currently there is a significant effort in both academia and industry for scalable machine learning, which also aligns with my interests. I believe most promising outcomes of data science involve a large amount of data, especially of very high dimensions. Therefore I care about developing novel and scalable algorithms that provably exploit the underlying problem structure such as sparsity or low-rank properties for big data. For deep learning models, I specifically also care about stabilizing their training process via designing better network architecture, as well as improving and understanding its adversarial robustness. I’m recently more interested in understanding the dynamics of minimax problems with applications of training deep generative models and adversarial training.