Label Efficient Learning with Explanations

State-of-the-art neural models have achieved impressive results on a range of NLP tasks but are still quite data hungry to learn. Training these models towards a specific task/domain may require hundreds of thousands of labeled samples, putting significant labor burden and time cost on manual data annotation. In this talk, going beyond the standard instance-label training design, I will share recent advances on learning with less labeled data by exploring high-level human supervision and knowledge. I will introduce neural rule grounding, a framework that enables softly grounding of surface pattern rules over unlabeled corpus to augment model training. Next, I will discuss how to extend the soft grounding framework to handle natural language explanation that are complex and compositional, and present neural execution tree, a modular network architecture for modeling the explanations. Lastly, I will briefly introduce a knowledge-aware graph network for incorporating multi-relational knowledge into pre-trained language models for commonsense question answering.

Xiang Ren is an assistant professor of Computer Science at USC, with affiliated appointment at USC ISI. He is the director of Intelligence and Knowledge Discovery (INK) Research Lab, the Information Director of ACM SIGKDD, and member of USC Machine Learning Center. Priorly, he was a research scholar at Stanford University, and received his Ph.D. in Computer Science from University of Illinois Urbana-Champaign. Dr. Ren’s research focuses on developing label-efficient, prior-informed computational methods that extract machine-actionable knowledge from natural-language data, as well as performing neural-symbolic reasoning over heterogeneous data. His research leads to a book and over 50 publications, was covered in over 10 conference tutorials, and received awards including faculty research awards from Google, Amazon, JP Morgan and Snapchat, ACM SIGKDD Dissertation Award, TheWebConference (WWW) Best Paper award honorable mention, and David J. Kuck Outstanding Thesis Award. He was named Forbes' Asia 30 Under 30 in 2019.