Talk Abstract
Commonsense knowledge is critical to achieving artificial general intelligence. This shared common background knowledge is implicit in all human communication, facilitating efficient information exchange and understanding. However, commonsense research is hampered by its immense quantity of knowledge because an explicit categorization is impossible. Furthermore, a plumber could repair a sink in a kitchen or a bathroom, indicating that common sense reveals a probable assumption rather than a definitive answer. To align with these properties of commonsense fundamentally, we want to model and evaluate such knowledge human-like using probabilistic abstractions and principles.
This talk will introduce a probabilistic model representing commonsense knowledge using a learned latent space of geometric embeddings -- probabilistic box embeddings. Using box embeddings makes it possible to handle commonsense queries with intersections, unions, and negations in a way similar to Venn diagram reasoning. Meanwhile, we show limitations with current Large Language models with their (commonsense) reasoning ability. Finally, I will discuss two benchmarks for evaluating commonsense in large language models. One includes a method to retrieve commonsense question-answer distributions from human annotators, and another focuses on assessing the long-tail (uncommon) part of commonsense knowledge. The combination of modeling and evaluation benchmarks sheds light on future commonsense research during LLMs.
Biography
Xiang Lorraine Li is an assistant professor in the Department of Computer Science at the University of Pittsburgh. She worked as a young investigator with the Mosaic team at AI2 before joining Pitt. Previously, she defended her Ph.D. in Computer Science from UMass Amherst in August 2022, working with Andrew McCallum. She obtained an M.S. in Computer Science from The University of Chicago while conducting research at TTIC. Her research is at the intersection of natural language processing, commonsense reasoning, knowledge representation, and machine learning. More specifically, her research focuses on designing probabilistic models and evaluation methods for implicit commonsense knowledge in language. One of her research publications was presented as a spotlight presentation (top 1.5%) at ICLR 2019. In addition, she regularly serves on program committees and workshop organizers in the NLP and ML fields, such as ICLR, ICML, NeurIPS, ACL, EMNLP, NAACL, ARR etc.
Location
Sennott Square Building, Room 5317
Date
Friday, November 17 at 2:00 p.m. to 3:15 p.m.