AI Seminar

Exploiting Domain Knowledge and Modular Semantics in Deep Learning for Natural Language Comprehension and Grounding

Parisa KordjamshidiAssistant ProfessorMichigan State University
3725 Beyster BuildingMap


Exploiting Domain Knowledge and Modular Semantics in Deep Learning for Natural Language Comprehension and Grounding


The recent research results in Natural Language Processing and many other complex problems show that deep learning models, especially, transformer-based language models, trained on large volumes of data suffer from a lack of interpretability and generalizability. While they might surprise us with writing an article that reads fluently given a prompt, they can easily disappoint us by failing in some basic reasoning skills like understanding that “left” is the opposite direction of “right”. For solving real-world problems, we often need computational models that involve multiple interdependent learners, along with significant levels of composition and reasoning based on additional knowledge beyond available data.  In my talk,  firstly, I will discuss our recent research on developing deep learning architectures for solving natural language comprehension and grounding in vision that  1) operate on capturing modular semantics from data, 2) capture high order patterns in the data that enable relational reasoning, 3) consider domain knowledge in learning.  Secondly, I will introduce our Declarative learning-based programming framework, DomiKnowS, that is designed to help in the integration of learning and reasoning and exploiting both symbolic and sub-symbolic representations for solving complex and AI-complete problems. With this framework domain knowledge represented symbolically in the form of constraints can be seamlessly integrated in deep models using various underlying algorithms.


Parisa Kordjamshidi is an assistant professor of Computer Science & Engineering at Michigan State University. Her research interests are machine learning and natural language processing. She has worked on the extraction of formal semantics and structured representations from natural language. She obtained an NSF CAREER award in 2019 to work on combining learning and reasoning for spatial language understanding. She is currently the leading PI of a project supported by Office of Naval Research to research on the integration of domain knowledge into statistical/neural learning. She is directing a research lab on Heterogeneous Learning and Reasoning (HLR) focusing on various NLP problems as well as combining vision and language. She obtained her Ph.D. from KU Leuven in 2013 and was a post-doc in University of Illinois at Urbana-Champaign until 2016. She was an assistant professor in Tulane University 2016-2019 before joining MSU. Kordjamshidi is a member of Editorial board of Journal of Artificial Intelligence Research (JAIR). She has published papers, organized international workshops and served as (senior) area chair/(senior) program committee of conferences including *ACL, EMNLP, IJCAI, AAAI, ECAI, ECML, COLING. She has contributed in organizing NAACL, EMNLP and ECML-PKDD conferences during the past few years.

Zoom (password: UMichAI)


AI Lab

Faculty Host

Joyce Chai