My research focus is to build more channels between SE and AI to make each community benefit more from the other. My past research has applied AI to SE, i.e., to use search-based techniques to automatically infer test oracles (ASE 2014), and to use machine learning techniques to automatically predicting the program behaviours (ISSTA 2016, TSE 2018). Currently, I am trying to combine my expertise in SE (e.g., code mutation and metamorphic testing) and AI (e.g., search algorithms) to improve automatic program debugging, as well as to apply SE methodology to AI to detect overfitting.