Manas Gaur

Biography

Manas Gaur is an assistant professor in the department of computer science and electrical engineering at the University of Maryland Baltimore County. He is an affiliate faculty member of Ebiquity Research Center at UMBC and directs the Knowledge-infused AI and Inference lab (https://www.csee.umbc.edu/research/research-labs/). As the AAAI New Faculty, his noted research in knowledge-infused learning explores methods and metrics to develop safe and explainable conversational systems. He has been an Eric and Wendy Schmidt DSSG Fellow with UChicago, AI for Social Good Fellow with Dataminr Inc. and Visiting Researcher at Alan Turing Institute.

Talks and Events

2023 Talk: Targeted Knowledge Infusion To Make Conversational AI Explainable and Safe

Conversational Systems (CSys) represent practical and tangible outcomes of advances in NLP and AI. CSys see continuous improvements through unsupervised training of large language models (LLMs) on a humongous amount of generic training data. However, when these CSys are suggested for use in domains like Mental Health, they fail to match the acceptable standards of clinical care, such as the clinical process in Patient Health Questionnaire (PHQ-9). The talk will present, Knowledge-infused Learning (KiL), a paradigm within NeuroSymbolic AI that focuses on making machine/deep learning models (i) learn over knowledge-enriched data, (ii) learn to follow guidelines in process-oriented tasks for safe and reasonable generation, and (iii) learn to leverage multiple contexts and stratified knowledge to yield user-level explanations. KiL established Knowledge-Intensive Language Understanding, a set of tasks for assessing safety, explainability, and conceptual flow in CSys.

Track: Deep Learning for and with Knowledge Graphs

Session Topics:

• Knowledge-infused Learning
• Process Knowledge
• Explainable AI
• Safe AI
• Conversational Systems
• Social Good

2022 Tutorial: Knowledge-infused Reinforcement Learning

Virtual health agents (VHAs) have received considerable attention, but the early focus has been on collecting data, helping patients follow generic health guidelines, and providing reminders for clinical appointments. While presenting the collected data and frequency of visits to the clinician is useful, further context and personalization are needed for a VHA to interpret and understand what the data means in clinical terms. This has made their use in managing health limited. Such understanding enables patient empowerment and self-appraisal — i.e., aiding the patient in interpreting the data to understand the changes in the patient’s health conditions, and self-management — i.e., to help a patient better manage their health through better adherence to the clinician guidelines and clinician recommended care plan. Crisis conditions such as the current pandemic have further stressed our healthcare system and have made the need for such advanced support more attractive and in demand. Consider the rapid growth in mental health because the patients who already had mental health conditions worsen, and many develop such conditions due to the challenges arising from lockdown, isolation, and economic hardships. The severe lack of timely availability of clinical expertise to meet the rapidly growing demand provides the motivation for advancing this research in developing more advanced VHAs and evaluating it in the context of mental health management.