I am working on Natural Language Processing (NLP). My current research focuses on three main themes:

1. Out-of-distribution generalization: domain adaptation, semi-supervised learning, learning with light supervision, cross-language and multilingual learning, zero-shot learning, multi-modal learning (language, vision, behavioral signals, etc.) and multi-task learning.

2. The intersection of computational language processing, human communication (e.g. conversations, social media), social, behavioral and health science (economics, game theory, psychology and psychiatry).

3. Causal inference for NLP. Causality research provides a bridge between the above two themes, with important applications to, e.g., model interpretation and out-of-distribution generalization.