Reading to Learn: Improving Generalization by Learning From Language

Talk by Victor ZHONG

January 31, 2023 Tuesday

Abstract:

Traditional machine learning (ML) systems are trained on vast quantities of annotated data or experience. These systems often do not generalize to new, related problems that emerge after training, such as conversing about new topics or interacting with new environments. In this talk, I present Reading to Learn, a new class of algorithms that improve generalization by learning to read language specifications, without requiring any actual experience or labeled examples. This includes, for example, reading FAQ documents to learn to answer new questions and reading manuals to learn to play new games. I will discuss new algorithms and data for Reading to Learn applied to a broad range of tasks, including pretraining for grounded reinforcement learning, data synthesis for code generation, and task-oriented dialogue about new topics, while also highlighting open challenges for this line of work.  Ultimately, the goal of Reading to Learn is to democratize AI by making it accessible for low-resource problems where the practitioner cannot obtain annotated data at scale, but can instead write language specifications that models read to generalize.

Speaker Bio:

Mr. Victor ZHONG

Ph.D. candidate, University of Washington

Victor Zhong is a PhD student at the University of Washington Natural Language Processing group. His research is at the intersection of natural language processing and machine learning, with an emphasis on how to use language understanding to learn more generally and more efficiently. His research covers a range of topics, including dialogue, code generation, question answering, and grounded reinforcement learning. Victor has been awarded the Apple AI/ML Fellowship as well as an EMNLP Outstanding Paper award. His work has been featured in Wired, MIT Technology Review, TechCrunch, VentureBeat, Fast Company, and Quanta Magazine. He was a founding member of Salesforce Research, and has previously worked at Meta AI Research and Google Brain. He obtained a Masters in Computer Science from Stanford University and a Bachelor of Applied Science in Computer Engineering from the University of Toronto.