I am interested in language interfaces that accommodate both the precise computer action space and the informal human thinking. Towards this, we studied learning language games where the system learns a language interactively starting from scratch; and naturalizing a programming language, where the system learns from hierarchical instructions.

More broadly, I’m interested in machine learning and natural language processing. Some of my work is applied: semantic parsing, machine translation, document classification and sentiment analysis. Others are more theoretical: method of moments, regularization/dropout, and MAP inference.