Assistant Professor, Tulane University
Thursday, February 21
11:15am - 12:30pm
The recent research results in Natural Language Understanding (NLU) and other problem domains show that monolithic deep learning models trained on merely large volumes of data suffer from lack of interpretability and generalizability. For NLU, 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. NLU requires pragmatics and common sense reasoning on top of syntactic and semantic information. Conventional programming paradigms offer no help in developing such complex learning-based models. In this talk, I discuss two themes of my research. One theme is the declarative learning-based programming (DeLBP) paradigm that aims at facilitating the design and development of complex intelligent systems. The other theme is an important NLU problem of spatial language understanding. Spatial language conveys the information about the location/translocation of objects and their spatial relationships. This semantics is relevant for visualization and grounding language into the real-world. I demonstrate how DeLBP framework facilitates working with structured data from heterogeneous resources (vision and language), considering domain knowledge and spatial ontologies in learning, and designing various learning and inference configurations. This paradigm helps to move towards integrating learning and reasoning and exploiting both symbolic and sub-symbolic representations for solving complex and AI-complete tasks.
Parisa Kordjamshidi is an assistant professor of computer science at Tulane University and holds a joint appointment as a research scientist at IHMC from July 2016. Her research interests are machine learning, natural language understanding, and declarative learning-based programming. She has worked on the extraction of formal semantics and structured representations from natural language, with a specific focus on spatial semantic representation and structured output learning models. She is the leading PI of a project funded by the Office of Naval Research (ONR) to perform basic research and develop a declarative learning-based programming framework for integration of domain knowledge into statistical learning (2019-2022) and received an NSF career award in 2019. She obtained her Ph.D. from KU Leuven in 2013 and was a post-doc in University of Illinois at Urbana-Champaign in cognitive computation group, before joining Tulane and IHMC. Parisa is a member of Editorial board of Journal of Artificial Intelligence Research (JAIR), a member of Editorial Board of Machine Learning and Artificial Intelligence, part of the journal of Frontiers in Artificial Intelligence and Frontiers in Big Data. She has published papers, organized several international workshops and served as a program committee member of conferences such as IJCAI, AAAI, ACL, EMNLP, COLING, ECAI and a member of organizing committee of NAACL-2018 and ECML-PKDD-2019 conferences.