Reasoning is one of the central topics in Artificial Intelligence. Informal reasoning in natural language is deeply entangled with semantics and is among the hardest problems to solve. In this talk, Dr. Zhu will discuss the work of his Lab on learning semantic representation and modelling reasoning in natural language, including efforts on semantic composition, natural language inference, and Winograd schema challenges. He will also present the Lab’s ongoing projects, using these models for various applications.