The Data Science Innovation Lecture Series features the following two guest speaker events:
Virtual Reality and Augmented Reality Applications for Situational Awareness and Emergency
Response using AI and Machine Learning
October 24, 2025 at 2pm
Presented by Dr. Sharad Sharma, Associate Dean of Research, Sinha Endowed Professor,
and Directory of Data Visualization and Extreme Reality Lab (DVXR). In the rapidly
evolving landscape of emergency response and decision making, the integration of Virtual
Reality (VR) and Augmented Reality (AR) technologies with Artificial Intelligence
(AI) and Machine Learning (ML) has opened new frontiers in enhancing situational awareness
and response strategies. VR offers a fully immersive environment that can simulate
emergency scenarios for training purposes such as active shooter events and fire evacuation
drills. VR-based emergency training modules allow first responders, security personnel,
and even civilians to engage in lifelike crisis situations without facing actual danger.
On the other hand, AR overlays digital information onto the real-world environment
through mobile devices or AR glasses. AR does not replace the real world but enhances
it by making it particularly useful during actual emergencies. Utilizing the visualization
and interaction capability that AR offers and the need for emergency evacuation training,
this talk explores mobile AR application (MARA) constructed to help users evacuate
a building in the event of emergencies such as a building fire, active shooter, earthquakes,
and similar circumstances. This talk also describes the AR application to leverage
the Microsoft HoloLens for emergency response during a building evacuation. This talk
also presents an immersive VR environment for performing virtual building evacuation
drills for active shooter and fire evacuation training scenarios using Meta Quest
head-mounted displays. It incorporates the use of avatars (user-controlled agents)
or agents (AI agents) may influence the engagements of the user experience for emergency
response, and training in emergency scenarios. The talk will also present the digital
twin research projects and Brain Computer Interface (BCI) projects in the DVXR lab.
These projects leverage state-of-the-art neurotechnology to better understand and
mitigate the impact of stress, anxiety, and depression. The integration of EEG data
from Galea and EMOTIV HMDs facilitate richer insights and potentially transform mental
health care through personalized, real-time interventions.
For more information on research projects: https://ci.unt.edu/dvxr/research/
Unifying Model, Data, and Innovation Evaluation for Trustworthy AI Testing
November 7, 2025 at 2pm
Discovery Park, Room B185
Presented by Dr. Junhua Ding, Reinburg Endowed Professor in Data Science and Chair
of the Anuradha & Vikas Sinha Department of Data Science at the University of North
Texas. A scholar, innovator, and leader, Dr. Ding’s work bridges the gap between academic
research and real-world impact. His research focuses on data-centric artificial intelligence,
biomedical computing, software security, and automated software engineering, and he
has authored more than 130 peer-reviewed publications in these areas and secured more
than $6 million in research grants from federal agencies and industry as a PI and
coPI.
This lecture introduces a systematic framework for testing and validating AI systems
by integrating model behavior analysis, data quality assessment, and innovation evaluation.
The presentation identifies critical gaps between traditional software testing and
AI model validation, emphasizing the need for metrics that ensure robustness, fairness,
and reliability.
The talk will present methods such as metamorphic testing, transfer learning evaluation,
and data-centric analysis to address dataset redundancy, bias, and domain shift. Through
a case study on medical concept normalization, it demonstrates how dataset overlap
can inflate performance metrics and proposes data quality evaluation and LLM-based
data augmentation to improve reliability.
Additionally, the talk explores formal verification of AI-generated artifacts, focusing
on prompt validation using model checking, few-shot learning, and temporal logic reasoning
to ensure specification correctness and sufficiency. Finally, it introduces a unified
perspective that connects model testing, data validation, and innovation measurement,
paving the way for more trustworthy and transparent AI systems.