Event box

Moderator:  Syed Ahmad Chan Bukhari, Collins College of Professional Studies, Computer Science, Mathematics and Science
Speaker:  Dr. Khalid Malik, Associate Professor at Oakland University 

Talk Summary:  Humans are capable of learning visual concepts by jointly understanding vision and language. Multiple modalities are also in play when making better diagnosis and treatment decisions after interpreting medical images in parallel to patient history, and when offering a personalized explanation to different stakeholders. However, state-of-the-art Deep Learning (DL) models do not take reasoning on multi-modal data into account, and instead rely heavily on a large volume of high-quality training data. Current DL models also do not exploit domain knowledge. They offer slow convergence and are not explainable by nature. Additionally, industrial data across the sites or organizations is non-independently and identically distributed (i.e., non-IID) and due to increasing privacy concerns and data regulations it is not possible to transfer information across organizations/sites. Thus, it is important to develop AI models which offer personalized explainability, allow human-in-loop AI, offer joint processing of multimodal data and apply reasoning to it, and offer context-aware distributed processing of data without the need to transfer it to a centralized location.  

By taking example of subarachnoid hemorrhage prediction, this talk will focus on explaining methods to extract features by jointly processing multi-modal data using context-aware, human-in-loop, privacy-preserved, federated DL approaches, and to then manipulate these features using a symbolic approach in order to offer higher accuracy, better explainability, and faster convergence than state-of-the-art deep learning models. 

Speaker’s Bio:  Dr. Khalid Malik is Associate Professor, in the department of Computer Science and Engineering at Oakland University. Dr. Malik’s research interests include trustworthy and decentralized neuro-symbolic artificial intelligence in cybersecurity and healthcare. In cybersecurity, he focuses on developing forensic examiners for authenticity, integrity and veracity of audio and videos by using explainable AI (XAI). He also works on hybrid cryptographic and generative AI models for secure group communication. In healthcare, he focuses on prediction of neurological disorders with focus on subarachnoid hemorrhage prediction, and infectious diseases using clinical text and medical imaging by using neruo-symbolic learning and automated knowledge graph generation, XAI and fairness-enabled AI, and federated learning. Dr. Malik’s research is supported by various international, federal, and state agencies such as the National Science Foundation, Brain Aneurysm Foundation, and Michigan Translational Research and Commercialization Innovation Hub. 

Please join us at 2:00 pm by clicking the Webinar linkhttps://sju.webex.com/sju/j.php?MTID=m074aa6589c3393a722790017cef74113 

Date:
Thursday, April 21, 2022
Time:
2:00pm - 3:00pm
Location:
Virtual
Campus:
Online
Categories:
  Research Insights  
Registration has closed.

Event Organizer

Profile photo of CTL Organizer
CTL Organizer