I have broad interests in machine learning, computer vision and natural language processing. My research objective is to build practical machine learning systems that can be more data efficient and generalisable for real-world applications. My current major research topics are about: low-supervision machine learning, including semi-supervised learning, unsupervised learning, few-shot/zero-shot learning etc; generalisable machine learning systems, including domain generalization, compositional generalization,etc.; applications in computer vision, e.g., dense prediction, fine-grained recognition, content generation; applications in natural language processing, e.g., low-resource NLP, generalization of NLP systems.