◆  Rada Mihalcea, University of Michigan

Keynote Topic: Towards People-centric Word Representations

Short Bio: Rada Mihalcea is the Janice M. Jenkins Collegiate Professor of Computer Science and Engineering at the University of Michigan and the Director of the Michigan Artificial Intelligence Lab. Her research interests are in computational linguistics, with a focus on lexical semantics, multilingual natural language processing, and computational social sciences. She serves or has served on the editorial boards of the Journals of Computational Linguistics, Language Resources and Evaluations, Natural Language Engineering, Journal of Artificial Intelligence Research, IEEE Transactions on Affective Computing, and Transactions of the Association for Computational Linguistics. She was a program co-chair for EMNLP 2009 and ACL 2011, and a general chair for NAACL 2015 and *SEM 2019. She currently serves as ACL President. She is the recipient of a Presidential Early Career Award for Scientists and Engineers awarded by President Obama (2009), an ACM Fellow (2019) and a AAAI Fellow (2021). In 2013, she was made an honorary citizen of her hometown of Cluj-Napoca, Romania.

Homepage: https://web.eecs.umich.edu/~mihalcea.

Abstract: Most of the work to date in natural language processing has relied on general purpose language representations, obtained from training one model on very large text collections. While this approach is effective for those people whose language style is well represented in the data, it quickly fails when it comes to the language spoken by those whose style diverges from the majority. In this talk, I will describe research that points to the value of people-centric language representations, and show that we can effectively use information about the people behind the words to build better natural language processing models.

◆  Yanchao Bi, Beijing Normal University

Keynote Topic: Semantic representation in the human brain

Short Bio: Yanchao Bi is a ChangJiang professor in IDG/McGovern Institute for Brain Research and the State Key Laboratory of Cognitive Neuroscience and Learning, at Beijing Normal University. She received her PhD from the Department of Psychology, Harvard University in 2006. Her current work focuses on the study of functional and neural architecture associated with semantic memory and language, using cognitive, neuropsychological and neuroimaging methods. She serves on the board of “Society of Neurobiology of Language”, the editorial board of Journals Elife, Cognition, Cognitive Neuropsychology, and Neurobiology of Language. She has won various awards, scholarships or recognitions such as “The National Science Fund for Distinguished Young Scholars”, “The National Science Fund for Excellent Young Scholars”, Sackler scholar of psychophysiology, Fulbright scholar, and “rising star” in the Observer by the American psychological association.

Homepage: http://bilab.bnu.edu.cn/index.html.

Abstract: Human brain stores tremendous amount of knowledge about this world, which is the foundation of language, thought, and reasoning. What’s the neural codes of semantic knowledge representation? Is the knowledge “roses are red” simply the memory trace of perceiving the color of roses, stored in the brain circuits within color-sensitive neurons? What about knowledge that is not directly perceived by senses, such as “freedom” or “rationality”? What’s their relationship with machine-based knowledge representation? I will present some work from my lab that addresses this issue using cognitive, neuroimaging, and neuropsychological methods with healthy subjects, individuals with sensory deprivation (blind and deaf) or with brain damage. The findings point to a highly distributed system incorporating two different types of information coding – one based on distributed sensory experiences (embodied) and one based on language (symbolic).

◆  Sebastian Riedel, University College London and Facebook AI Research

Keynote Topic: Parametric vs Nonparametric Knowledge, and what we can learn from Knowledge Bases

Short Bio: Sebastian Riedel is a researcher at Facebook AI research, professor in Natural Language Processing and Machine Learning at the University College London (UCL) and an Allen Distinguished Investigator. He works in the intersection of Natural Language Processing and Machine Learning, and focuses on teaching machines how to read and reason. He was educated in Hamburg-Harburg (Dipl. Ing) and Edinburgh (MSc., PhD), and worked at the University of Massachusetts Amherst and Tokyo University before joining UCL.

Homepage: http://www0.cs.ucl.ac.uk/people/S.Riedel.html.

Abstract: Traditionally, AI and Machine Learning communities have considered knowledge from the perspective of discrete vs continuous representations, knowledge bases (KBs) vs dense vectors or logic vs algebra. While these are important dichotomies, in this talk I will argue that we should put more focus on another: parametric vs non-parametric modelling. Roughly, in the former a fixed set of parameters is used, in the latter parameters grow with data. I will explain recent advances in knowledge intensive NLP from this perspective, show the benefit of hybrid approaches, and discuss KBs as non-parametric approaches with relatively crude assumptions about what future information needs will be. By replacing these assumptions with a learnt model, we show that such “modern KBs” are a very attractive alternative or complement to current approaches.

◆  Graham Neubig, Carnegie Mellon University

Keynote Topic: How Can We Know What and When Language Models Know?

Short Bio: Prof. Graham Neubig is an associate professor at the Language Technologies Institute of Carnegie Mellon University. His work focuses on natural language processing, specifically multi-lingual models that work in many different languages, and natural language interfaces that allow humans to communicate with computers in their own language. Much of this work relies on machine learning, and he is also active in developing methods and algorithms for machine learning over natural language data. He publishes regularly in the top venues in natural language processing, machine learning, and speech, and his work has won awards at EMNLP 2016, EACL 2017, NAACL 2019, and ACL 2021.

Homepage: http://www.phontron.com.

Abstract: One recent remarkable finding in natural language processing is that by training a model to simply predict words in a sentence, language models can learn a significant amount of world knowledge. In this presentation, I will discuss a new paradigm of "prompting", which attempts to elicit knowledge from language models by presenting a textual prompt such as "CMU is located in ___" and asking the language model to fill in the answer. I will first give an outline of this paradigm as a whole, then present research regarding two questions. First: how can we most effectively elicit this knowledge from language models in this way, and what are the connections to other methods for parameter-efficient training of neural NLP models? Second: how can we best know when these predictions are accurate, and when they are no better than a random guess?