The central goal of my research is to make Artificial Intelligence (AI) systems more and more usable. An AI system includes two critical functions -- 1) perceiving the real world that AI interacts with and 2) expressing the personal characteristic/action and situated environment of AI itself. To ultimately achieve the usability of AI systems, my research focuses on the following two key directions -- 1) learning to perceive the real world to make the perceiving more effective with less supervision and 2) learning to generate/re-create the digital world to make the expressing easier with less human intervention. Further, my research also involves the construction of datasets and open-source software to facilitate the development of academia, centered around these two research directions. Moving forward, I am recently passionate about human-like AI systems., which will have better common sense in their perceiving system, as well as better creativity in their expressing system. This page lists some of my previous work.