Reposted from the Faculty of Engineering.
John and Mercia Price Engineering from the University of Utah held Utah's first AI Summit on June 18th, with artificial intelligence researchers from all departments gathering for the full-day symposium. They connected with over 400 faculty members, students, industry partners and policy makers across the region.
Charles Musgrave, dean of Price College of Engineering, has begun the sold-out process held at the SJ Quinney College of Law. As a chemical engineer, Musgrave has used machine learning technology for over a decade to pursue the development of new materials. But from the advantages of the dean and the cliffs of a new technological revolution, the outlook for artificial intelligence is endless.
“Those who lead AI lead in science, economics, national security and innovation,” Musgrave said. “But if we do it right, we will lead art, entertainment, and personal fulfillment.”
U Trustee Steven Price detailed how university research can return to the benefit of its citizens over the past decades. “These are the ingredients that fertile Utah's growth,” he said. “We're in motion and in moments. Movement is AI, and that moment is now. AI is moving rapidly, and we need to move faster.”
In addition to a series of poster presentations by more than 60 students from across the state, the symposium was organized into four panels. Each panel was booked by a series of one-minute “lightning talks” from the selection of student poster presenters and a Q&A session with viewers.
AI that senses, sees, and secures the world
Weilu GaoFaculty of Electrical and Computer Engineering, “Machine Learning of Optics”
Ziad al-HalahKahlert School of Computing, “Multimodal Evolutionary AI”
Guang Hong TaoKahlert School of Computing, “Towards a Large-scale Language Model for Safe and Secure”
Moderator Varun ShankarAssistant Professor at Kahlert School of Computing
Panelists engaged in the way AI systems are increasingly integrated with physical systems. Therefore, these systems need new ways to detect environmental features and ensure that they interact in ways that do not harm their people. Weilu Gao's work adds another dimension to this field. He and his colleagues recently said,Optical Neural EngineIt could potentially accelerate the calculations contained in these applications.
Next-generation AI: From supervision to autonomy
Jacob HochhalterFaculty of Mechanical Engineering, “HyperComplex Automated Differentiation with Derivative-Informed Data”
Daniel BrownKahlert School of Computing, “Towards a robust, interactive, and human-friendly AI system.”
Vivek SrikumarKahlert School of Computing, “Human Language Technology: Can we make it bigger?”
Moderator Tucker HermanAssociate Professor at Kahlert School of Computing
This panel examined the trade-offs that AI systems must make with regard to the data they are being trained to. While many machine learning techniques rely on large data sets, not all applications are accessible for practical, economic, or even legal reasons. As well as reducing the cost of upfront training, new strategies for building these models feature applications in low-resource and off-grid environments.
The bond between health, humans and machines
Torga TasdizenFaculty of Electrical and Computer Engineering, “Towards an Interpretable AI Model of Radiology.”
Neda Netaghdepartment Electrical and Computer Engineering“Towards less artificial intelligence”
Alan KunzKahlert School of Computing, “Autonomic Surgery Robots Learning from Human Surgeons”
Amir ArzaniFaculty of Mechanical Engineering, “Science Machine Learning: From No Data to Large Data”
Ashley DallimpleFaculty of Biomedical Engineering, “Reinforcement Learning to Predict Walking-Related Events”
Moderator Laura HalockAssistant Professor in the Faculty of Mechanical Engineering
This panel continued the theme of integration of AI into physical systems, in this case biological systems. Healthcare applications hold some of the best promises of AI systems, but there are plenty of technical and ethical issues. Panelists discussed automated analysis of medical imaging, visual neuroscience, physics-based models of blood flow and tumor tissue, and gait biomechanics.
“Bridge” AI and infrastructure
Ryan JohnsonFaculty of Civil and Environmental Engineering, “High-resolution snow mapping with machine learning: a pioneering product to enhance interseason water supply forecasts.”
Masood Parvaniadepartment Electrical and Computer Engineering“AI for automatic power grid operation”
Chenxi (Dylan) LiuFaculty of Civil Engineering and Environmental Engineering, “Utilizing AI in intelligent transport systems”
Taylor SparksFaculty of Materials Science and Engineering, “Today's Materials”
Moderator Kathy LiuAssociate Professor, Faculty of Civil Engineering and Environmental Engineering
Participants zoomed out to even larger physical systems, such as transport networks and other civic infrastructure. Here, the complexity of the large number of inputs and their interactions naturally fits machine learning techniques. Panelists discussed training against cyberattacks, the relationship between winter snowfall and summer water levels, and the digital twins in the electric grid to prevent crashes in chain reactions on highways, and how the materials that make up everyday life are currently invented with these technologies.
Media and PR contact information
-
Evan Lerner
Communications Director John and Mercia Price University of Engineering
801-581-5911
evan.lerner@utah.edu
