BEGIN:VCALENDAR VERSION:2.0 PRODID:IEEE vTools.Events//EN CALSCALE:GREGORIAN BEGIN:VTIMEZONE TZID:America/New_York BEGIN:DAYLIGHT DTSTART:20260308T030000 TZOFFSETFROM:-0500 TZOFFSETTO:-0400 RRULE:FREQ=YEARLY;BYDAY=2SU;BYMONTH=3 TZNAME:EDT END:DAYLIGHT BEGIN:STANDARD DTSTART:20251102T010000 TZOFFSETFROM:-0400 TZOFFSETTO:-0500 RRULE:FREQ=YEARLY;BYDAY=1SU;BYMONTH=11 TZNAME:EST END:STANDARD END:VTIMEZONE BEGIN:VEVENT DTSTAMP:20260210T165232Z UID:5B958C42-6416-4641-A95D-FDFF906413FD DTSTART;TZID=America/New_York:20260213T120000 DTEND;TZID=America/New_York:20260213T130000 DESCRIPTION:Title: Multimodal Knowledge Generation through Team Data Scienc e\n\nAbstract:\n\nIn this presentation I will discuss my experience workin g with diverse teams to advance human-in-the-loop computing capabilities o n multimodal data. I will review use cases and challenges using large repo sitories of text\, genomic\, sensor\, and imaging data in secure environme nts.\n\nBio:\nDr. Ioana Danciu is a research scientist and group leader at the Oak Ridge National Laboratory (ORNL). Her research focuses on adaptin g scalable computational methods using multimodal data (imaging\, text\, s ensor\, and omics) for domain science use cases. Her interest areas are do main-specific machine learning\, interpretable AI\, large scale computatio n\, learning with less labeled data\, human-computer interaction and healt hcare privacy and security. Before joining ORNL in 2018\, she worked as an engineer and researcher at Vanderbilt University and Vanderbilt Universit y Medical Center in Nashville\, TN for over a decade.\n\nTitle: Reliable A I for Safety-Critical Applications\n\nAbstract:\n\nThe increasing adoption of large language models (LLMs) and other transformer-based systems in sa fety-critical domains such as national security and healthcare raises fund amental challenges related to reliability\, robustness\, and trustworthine ss. This talk will present some of her recent work on developing reliable AI methodologies for safety-critical applications\, focusing on the use of LLMs for information extraction\, question-answering\, and classification \, where hallucinations and miscalibrated confidence pose significant risk s. The presentation will discuss mitigation strategies for improving factu al consistency\, confidence calibration\, and error detection\, as well as retrieval-augmented generation (RAG) architectures for grounding model ou tputs in external evidence and the associated challenges in evaluating RAG systems. Beyond nominal settings\, the talk will explore adversarial robu stness issues\, including data poisoning\, adversarial patch generation us ing LLMs\, and jailbreak attacks that expose critical failure modes in rea l-world deployment. Finally\, the presentation will briefly address interp retable predictive modeling approaches that combine transformer architectu res with attribution methods to support transparency and accountability\, positioning reliability as a multi-faceted technical challenge for deployi ng AI in safety-critical environments.\n\nBio:\n\nDr. Maria Mahbub is a Re search Associate in the Cyber Resilience and Intelligence Division of the National Security Sciences Directorate at ORNL. She earned her PhD in Comp uter Science from the University of Tennessee\, Knoxville\, in August 2023 . At ORNL\, she leads research in multimodal AI\, large language models\, and natural language processing for high-stakes applications. Her work foc uses on developing modern AI systems and strengthening their reliability i n deployment\, combining expertise in information extraction\, interpretab le predictive modeling\, and retrieval-augmented generation. She works ext ensively with multimodal data\, including structured records\, unstructure d text\, and genetic information\, across national security and bioinforma tics applications. Her research also spans adversarial robustness and comp uter vision\, with a focus on model behavior in real-world settings. She h as received the People's Choice Award as well as Second Place in the FY25 Your Science in a Nutshell Competition\, and she is an awardee of the FY26 LDRD Early Career Competition. She aims to advance AI systems that are tr ustworthy\, robust\, and empirically grounded.\n\nSpeaker(s): Ioana\, Mari a\n\nVirtual: https://events.vtools.ieee.org/m/538382 LOCATION:Virtual: https://events.vtools.ieee.org/m/538382 ORGANIZER:kolivera@ieee.org SEQUENCE:2 SUMMARY:IEEE WIE Virtual Seminar URL;VALUE=URI:https://events.vtools.ieee.org/m/538382 X-ALT-DESC:Description: <br /><p><strong>Title: </strong>Multimodal Knowled ge Generation through Team Data Science</p>\n<p><strong>Abstract:</strong> </p>\n<p><em>In this presentation I will discuss my experience working wit h diverse teams to advance human-in-the-loop computing capabilities on mul timodal data. I will review use cases and challenges using large repositor ies of text\, genomic\, sensor\, and imaging data in secure environments.< /em></p>\n<p><strong>Bio</strong>:<br><em>Dr. Ioana Danciu is a research s cientist and group leader at the Oak Ridge National Laboratory (ORNL). Her research focuses on adapting scalable computational methods using multimo dal data (imaging\, text\, sensor\, and omics) for&nbsp\;domain science us e cases. Her interest areas are domain-specific machine learning\, interpr etable AI\, large scale computation\, learning with less labeled data\, hu man-computer interaction and healthcare privacy and security. Before joini ng ORNL in 2018\, she worked as an engineer and researcher at Vanderbilt U niversity and Vanderbilt University Medical Center in Nashville\, TN for o ver a decade.</em></p>\n<p>&nbsp\;</p>\n<p><strong>Title:</strong> Reliabl e AI for Safety-Critical Applications</p>\n<p><strong>Abstract:</strong><s pan class="Apple-converted-space">&nbsp\;</span></p>\n<p>The increasing ad option of large language models (LLMs) and other transformer-based systems in safety-critical domains such as national security and healthcare raise s fundamental challenges related to reliability\, robustness\, and trustwo rthiness. This talk will present some of her recent work on developing rel iable AI methodologies for safety-critical applications\, focusing on the use of LLMs for information extraction\, question-answering\, and classifi cation\, where hallucinations and miscalibrated confidence pose significan t risks. The presentation will discuss mitigation strategies for improving factual consistency\, confidence calibration\, and error detection\, as w ell as retrieval-augmented generation (RAG) architectures for grounding mo del outputs in external evidence and the associated challenges in evaluati ng RAG systems. Beyond nominal settings\, the talk will explore adversaria l robustness issues\, including data poisoning\, adversarial patch generat ion using LLMs\, and jailbreak attacks that expose critical failure modes in real-world deployment. Finally\, the presentation will briefly address interpretable predictive modeling approaches that combine transformer arch itectures with attribution methods to support transparency and accountabil ity\, positioning reliability as a multi-faceted technical challenge for d eploying AI in safety-critical environments.</p>\n<p><strong>Bio</strong>: <span class="Apple-converted-space">&nbsp\;</span></p>\n<p>Dr. Maria Mahbu b is a Research Associate in the Cyber Resilience and Intelligence Divisio n of the National Security Sciences Directorate at ORNL. She earned her Ph D in Computer Science from the University of Tennessee\, Knoxville\, in Au gust 2023. At ORNL\, she leads research in multimodal AI\, large language models\, and natural language processing for high-stakes applications. Her work focuses on developing modern AI systems and strengthening their reli ability in deployment\, combining expertise in information extraction\, in terpretable predictive modeling\, and retrieval-augmented generation. She works extensively with multimodal data\, including structured records\, un structured text\, and genetic information\, across national security and b ioinformatics applications. Her research also spans adversarial robustness and computer vision\, with a focus on model behavior in real-world settin gs. She has received the People's Choice Award as well as Second Place in the FY25 Your Science in a Nutshell Competition\, and she is an awardee of the FY26 LDRD Early Career Competition. She aims to advance AI systems th at are trustworthy\, robust\, and empirically grounded.</p> END:VEVENT END:VCALENDAR