My research interests lie in machine learning, data mining, and artificial intelligence, with a particular emphasis on

Theoretical Level: Analysis of the generalization ability of learning algorithms; Theoretical understandings of various loss functions; Development of unbiased estimators for classification risk.

Algorithmic Level: Development of effective learning algorithms for various weakly supervised or complex learning frameworks: partial-label learning, noisy-label learning, positive-unlabeled learning, complementary-label learning, semi-supervised learning, multi-label learning, learning from crowds, learning with similarity information, unsupervised domain adaptation, etc.

Appliction Level: Recommender systems with implicit feedback.