My research interests lie in the broad area of statistical machine learning — a research field that addresses the statistical and computational properties of machine learning algorithms and their optimality guarantees. Specifically, my work focuses on developing provable and practical methods for various challenging learning regimes (e.g., high dimensional, heterogeneous, privacy-constrained, sequential, parallel and distributed) and often involves exploiting hidden structures in data (generalized sparsity, union-of-subspace, graph or network structures), balancing various resources (model complexity, statistical power and privacy budgets) as well as developing scalable optimization tools (e.g., those tailored for deep learning).

I am also interested in applications of statistics and machine learning, such as those in clean energy, health care, housing, financial market, web services and so on. The key challenges of many such problems are in fact about how we can effectively and efficiently use the available data to make sensible predictions (supervised learning), quantify uncertainty (statistical inference), design sequential experiments (active learning / bandits) and to infer long-term consequences of a sequence of decisions (reinforcement learning).