Teresa M. Donnellan

Brainpower inspires energy-efficient AI

Teresa Donnellan — Thu, 22 Jan 2026 16:57:20 +0000

Brainpower inspires energy-efficient AI Teresa Donnellan Thu, 01/22/2026 - 11:57

The key to energy-efficient artificial intelligence (AI) could lie in the human brain. �鶹��Ƶ researchers Maryam Parsa (PI) and Giorgio Ascoli (co-PI) are leading a Department of Energy-funded project to make AI dramatically more efficient by mimicking the brain’s computing strategies.

Maryam Parsa. Photo provided.

“AI today is powerful but extremely energy-hungry,” said Parsa, an assistant professor in George �鶹��Ƶ’s Department of Electrical and Computer Engineering who specializes in neuromorphic computing. “We want to understand why the human brain is so efficient and why AI is not—and then bridge that gap.”

Unlike traditional AI models that rely on continuous signals and massive data centers, the brain uses sparse, event-driven signals called spikes.

“Any one neuron only fires a spike here and there, but the timing of those spikes carries a lot of meaning,” explained Ascoli, a Distinguished University Professor in George �鶹��Ƶ’s Bioengineering Department specializing in neuroscience. “That’s why the brain can do so much with so little energy.”

To replicate this efficiency holistically, Parsa and Ascoli teamed up with Akhilesh Jaiswal at the University of Wisconsin–Madison and Pedram Khalili at Northwestern University. The team is building a full-stack neuromorphic computing system spanning devices, circuits, architectures, algorithms, and applications.

Each principal investigator is contributing specific expertise to the system:

Ascoli contributes neuroscience insights using Izhikevich models, mathematical frameworks that capture real neuron spiking dynamics, to inform system design.
Khalili develops spintronic devices, components that utilize magnetic materials to harness an electron's quantum properties of spin and charge for processing and storing data. By naturally exhibiting neuron-like temporal behavior, these devices are expected to reduce energy demands.
Jaiswal designs analog circuits to integrate these spintronic devices into functional architectures.
Parsa leads the project as well as neuromorphic learning algorithms and application-level integration, exploring diverse learning approaches to optimize speed, precision, resiliency, privacy, and energy savings.

Giorgio Ascoli. Photo provided.

“Izhikevich models allow neuromorphic systems to mimic the timing, diversity, and variability of real neurons, enabling more brain-like learning and efficiency,” Ascoli said. His extensive experience with these models, including simulating hippocampal networks, provides the biological foundation for the project.

“Instead of forcing existing devices to mimic brain behavior, we asked: What if we leverage new devices that naturally exhibit these dynamics?” Parsa said. This approach ensures innovations at the hardware level align with biologically inspired algorithms, creating a cohesive system rather than isolated solutions.

By introducing greater biological complexity and leveraging novel spintronic devices, the team hopes to reduce AI’s energy footprint while improving performance metrics such as accuracy and privacy. Early results are promising; even simple changes to spike distributions improved privacy without sacrificing accuracy.

The research also explores learning algorithms beyond the traditional. “Our goal is to see how different neuron models and learning rules interact and which combinations optimize for speed, precision, or energy savings,” Parsa said.

Ultimately, the project could transform AI by making it greener and more brain-like. “If we succeed, we’ll have systems that learn faster, consume less power, and operate more like the brain,” said Ascoli.

“This is not just about making AI more efficient,” said Parsa. “It’s about rethinking computing from the ground up."

In This Story

Maryam Parsa

Giorgio Ascoli

Explore AI at George �鶹��Ƶ

Topics

Bioengineering

Electrical and Computer Engineering

Research

Artificial Intelligence

Department of Energy

George �鶹��Ƶ’s Imaging and Vision Lab illuminates the future of digital production

Teresa Donnellan — Mon, 05 May 2025 19:57:46 +0000

George �鶹��Ƶ’s Imaging and Vision Lab illuminates the future of digital production Teresa Donnellan Mon, 05/05/2025 - 15:57

In This Story

At the intersection of imaging, computer graphics, and computer vision, �鶹��Ƶ’s Imaging and Vision Lab is pushing the boundaries of visual technology. Directed by Jinwei Ye, an associate professor in the Department of Computer Science, the lab explores every stage of the imaging pipeline—from data acquisition to 3D rendering—by designing novel algorithms, imaging systems, and immersive environments.

The lab, which is in Van Metre Hall at �鶹��Ƶ Square, has two new standout technologies in its arsenal: the Light Stage and the Pixel Cube.

The Light Stage—a spherical structure with around 150 LED lights pointing inward along with cameras—uses polarized, Wi-Fi-controlled lighting to create complex illumination patterns. This enables the lab to capture highly detailed 3D models of real-world objects, a capability that’s particularly useful in film, gaming, and robotics.

A student experimenting with Light Stage. Photo by Ron Aira/Office of University Branding

“We put objects in the center and take pictures under different lighting to recover really high-quality 3D models of real physical objects,” said Yufan Zhang, a PhD student working on this project. The Light Stage was fully designed and built by Ye’s students, providing them with a unique, hands-on engineering experience.

The Pixel Cube takes immersion a step further. Comprising 90 LED panels arranged in a cube, the system allows for panoramic simulations—perfect for applications in virtual production such as driving simulators.

“A very intuitive application of it is immersive viewing like a VR experience without using the headset,” said Ye.

But the lab’s ambitions stretch beyond visual hardware. Ye’s team is building a large dataset of everyday objects, intended to support generative AI development for uses ranging from entertainment to robotic training.

A student in the Pixel Cube. Photo by Ron Aira/Office of University Branding

“We’re creating a large 3D model dataset of diverse real-world objects, which can be used as assets either in films or in games or shown to robots,” Ye said.

Collaboration is key to the lab’s impact, added Ye. Her team, focused on advancing computational photography and AI-integrated imaging workflows, remains committed to developing foundational tools for tomorrow’s imaging technologies. They work closely with other George �鶹��Ƶ faculty and industry partners and invite potential collaborators to reach out about interdisciplinary projects in fields such as film studies, video game development, creative visual arts, simulation, and virtual training.

“We really want to make full use of this [equipment], so we want as many people as possible to use it,” said Ye.

With its combination of advanced equipment, collaborative ethos, and dedication to hands-on student involvement, the Imaging and Vision Lab is making George �鶹��Ƶ a destination for next-generation research in computer vision and graphics.

Topics

Innovative approaches propel STAT 250 to new heights at �鶹��Ƶ

Teresa Donnellan — Mon, 31 Mar 2025 14:40:51 +0000

Innovative approaches propel STAT 250 to new heights at �鶹��Ƶ Teresa Donnellan Mon, 03/31/2025 - 10:40

In This Story

Kenneth Strazzeri

Many university students taking a general education statistics course might be tempted to shirk attendance. But �鶹��Ƶ’s STAT 250 Introductory Statistics I boasts an impressive attendance rate of 80% across its sections. With between eight and ten sections per semester, the course is offered in-person and online both synchronously and asynchronously. It serves as one of the most popular quantitative reasoning courses in �鶹��Ƶ Core, the university's general education program, with enrollments of nearly one thousand students per semester.

The course’s success can be attributed in part to Associate Professor Kenneth Strazzeri, the course coordinator. Strazzeri tweaks the curriculum each semester to optimize student engagement.

Kenneth Strazzeri, Photo by Office of University Branding

“I want students to succeed. That's been my number one goal since taking this type of role at �鶹��Ƶ,” said Strazzeri. “When exposed to statistics again, I want them to think back and say, ‘Oh, I remember that. I took stat, and I had a good handle on these basic concepts.’”

Strazzeri’s rich background in statistics education makes him ideal for his role. He holds a PhD in mathematics education and an MS in statistics from the University of Virginia.

In STAT 250, Strazzeri focuses on increasing student engagement through various motivational strategies. Attendance bonuses and small incentives for completing course evaluations were introduced to encourage students to actively participate. Another key improvement was the restructuring of assignments. Previously, students were required to complete three major assignments, but these have now been split into six smaller assignments, a change that allows for more frequent feedback and helps students stay on track throughout the semester.

One course innovation that excites Strazzeri has been the introduction of simulation-based methods to teach statistical inference, a process used to draw conclusions about a population based on a sample of data taken from that population.

"We are teaching inference at an informal level first to then get to probability and modeling," Strazzeri explained.

This approach helps students grasp complex concepts related to statistical inference such as the interpretation of a confidence interval by using simulation methods such as bootstrapping—a technique used to estimate the variation of a sample statistic using only the original sample selected—at an informal level before delving into more formal methods.

Looking ahead, Strazzeri is excited about the future of STAT 250, particularly with the transition to the Canvas learning management system. He is working with Instructional Designer Abigail Smith in George �鶹��Ƶ’s Stearns Center for Teaching and Learning to make the course even more interactive and user-friendly.

"The future of STAT 250 is bright because of the new design in Canvas," Strazzeri said. Their plans include embedding questions in instructional videos and using those questions live during in-person classes to enhance student interaction and understanding.

Overall, the efforts led by Strazzeri have significantly improved the learning experience in STAT 250. By incorporating modern teaching methods, restructuring assignments, and increasing student engagement, the Department of Statistics is ensuring students not only succeed in the course but also develop a strong foundation in statistical concepts.

Topics

Department of Statistics

Cyber sweep: George �鶹��Ƶ students take top three spots in national transportation cybersecurity competition

Teresa Donnellan — Thu, 27 Mar 2025 14:42:35 +0000

Cyber sweep: George �鶹��Ƶ students take top three spots in national transportation cybersecurity competition Teresa Donnellan Thu, 03/27/2025 - 10:42

"I'm going to teach you how to hack a modern car," said �鶹��Ƶ professor Tanvir Arafin to the fall 2024 cohort of CYSE 465 Transportation Systems Design. Arafin not only introduced his students to such exciting concepts in transportation security, but he also incorporated the opportunity to participate in a nationwide cybersecurity competition into his course.

The entire class of 12 from the Department of Cyber Security Engineering chose to participate in the optional assignment and several won first, second, and third place at the National Transportation Cybersecurity Competition (NTCC) in November 2024. George �鶹��Ƶ’s Finn Schaefer won first place; the team of Noor Mohammed, Sonia Shaukat, Caleb Hughston, and Kayleigh Batchos won second place; and Navraj Singh Gill won third place.

Adobe Stock

The national competition was open to all university students, both undergraduate and graduate, who could participate individually or in teams of up to four, guided by a faculty advisor. This year’s competition was titled “CYBER-CARE Challenge 2024: Securing the Future of Transportation.”

Participants faced a series of cybersecurity challenges, including password cracking, traffic management system exploitation, Vigenère cipher decryption, SQL injection vulnerability identification, cross-site scripting (XSS) attacks, and vehicle trajectory anomaly detection. They had a few weeks to complete the challenges remotely.

Arafin developed CYSE 465 by integrating real-world scenarios and cutting-edge research findings from his lab. He has been working in hardware and cyber-physical systems security with a key focus on cybersecurity issues of autonomous and transportation systems. CYSE 465 covers understanding data, communication, operating systems, control, and security primitives used in today's vehicles for safe and intelligent control and navigation.

“The course offers a unique perspective on the cyber vulnerabilities that emerge from the security-oblivious modernization of transportation systems,” said Arafin.

While he has taught the course since 2022, this was the first time he incorporated the competition into the course. Arafin noted that such competition participation can greatly enhance students’ resumes, and first-place winner Schaefer, a busy student, greatly appreciated the chance to participate.

“This was actually the first competition that I've competed in,” he said, explaining that it’s hard to find time for such activity outside of school and work. Schaefer’s favorite challenge at the NTCC involved assessing vehicle datasets to detect problems or attacks. He drew from his personal interest in machine learning to solve the challenge.

“I've worked with machine learning models a lot in my own personal projects, but I've never had an applicable use for it,” said Schaefer. “So, it was interesting to actually see my work have real meaning to it for once. I was able to do something pretty cool.”

Through courses like CYSE 465, which bridge the gap between theoretical knowledge and practical application, the Department of Cyber Security Engineering is fostering a new generation of cybersecurity professionals who are well-equipped to secure the future of transportation systems.

"The students’ competition performance is a very compelling argument on the benefits of bringing the results and lessons learned in the research lab directly to the classroom, something many CYSE faculty are doing," said Paulo Costa, the department's faculty chair.

The achievements of George �鶹��Ƶ students at the NTCC highlight the importance of this approach. By bringing state-of-the-art research directly to the classroom, the Department of Cyber Security Engineering is preparing students to become leaders in the field. As the cybersecurity landscape continues to evolve, the department's innovative educational strategies will ensure that its graduates are ready to meet the challenges of tomorrow.

In This Story

Tanvir Arafin

Paulo Costa

Learn more about Cybersecurity at George �鶹��Ƶ

Topics

Cybersecurity

transportation engineering

Student Success

Department of Cyber Security Engineering

Students

innovative classes

Strategic Direction

Nearly 500 high schoolers get expert advice at Virginia CyberSlam 2025

Teresa Donnellan — Mon, 17 Mar 2025 16:20:56 +0000

Nearly 500 high schoolers get expert advice at Virginia CyberSlam 2025 Teresa Donnellan Mon, 03/17/2025 - 12:20

Nearly 500 high school students from across the commonwealth gathered at �鶹��Ƶ for Virginia CyberSlam 2025, where cybersecurity experts from both the public and private sectors shared industry insights during a day of activities aimed at inspiring the next generation of cybersecurity professionals.

Dolores Garza addresses CyberSlam 2025 attendees. Photo by Teresa Donnellan/CEC

“What an event it was March 10, with 475 students representing 22 schools and six school districts from around the commonwealth enjoying hands-on workshops and activities led by cyber experts from industry, government, and higher education,” said CyberSlam founder Chris Starke, a teacher for Loudoun County Public Schools who co-directed the event with the University of Alabama in Huntsville’s Jennifer Marden. “In all, 28 presenters combined to run 12 workshops—6 for each of our novice and advanced students. It was truly rewarding to see students enjoy learning about cybersecurity.”

Students heard from keynote speaker Dolores Garza, Branch Chief of the Technical Operations Group at the United States Marshals Service, among others. Garza oversaw the creation of a cyber investigation unit at the U.S. Marshals and described the evolution of cybersecurity as a professional field. She encouraged attendees to stay open to new technological developments and emphasized the need for a hybrid set of skills that combines criminal investigatory techniques with technical expertise.

“Diversity in technology is not just an asset; it is a necessity,” said Garza. “The more perspectives we bring to problem-solving, the stronger and more effective our solutions become.”

Student navigates drone during CyberSlam 2025. Photo by Ron Aira/Office of University Branding.

Throughout the day, professionals led hands-on workshops in various aspects of cybersecurity.

David Raymond from Virginia Cyber Range led a session on advanced Capture the Flag challenges, focusing on Wireshark.

Greg Crabb from 10-8, LLC, conducted a tabletop exercise on incident response and deep fake technology, and Randy Benn from Claroty introduced students to the Internet of Things.

Other experts leading workshops represented Vertex11, OpusType Advisors, the Idaho National Laboratory, the Drug Enforcement Administration, CYBER.ORG, George �鶹��Ƶ, and Shenandoah University, leading workshops on drones, deep fakes, digital forensics, and more.

Students participate in a CyberSlam workshop. Photo by Ron Aira/Office of University Branding

“I would like to thank our co-hosts, George �鶹��Ƶ’s Department of Cyber Security Engineering and Loudoun County Public Schools, as well as the Commonwealth Cyber Initiative Northern Virginia Node (CCI NoVA Node), Virginia Cyber Range, Vertex11, CACI International Inc., Claroty, CodeHS, Widelity, Inc., and GT Edge AI for sponsoring this wonderful event that in four years has impacted more than 1,600 students,” said Starke. “And, thank you to our exhibitors the Cybersecurity and Infrastructure Security Agency, �鶹��Ƶ Competitive Cyber (MCC), George �鶹��Ƶ’s College of Engineering and Computing, GT Edge AI, and the Virginia National Guard 91^st Cyber Brigade.”

CyberSlam attendees navigate drones in a workshop. Photo by Jennifer Marden

One event highlight was a panel featuring George �鶹��Ƶ students from the MCC club—Dylan Charnick, Dylan Knoff, and Danyaal Shaozab. MCC is known for its inclusivity, welcoming George �鶹��Ƶ students from all majors and levels of experience. The club meets twice weekly, hosting impressive speakers, running workshops on cybersecurity skills, and forming teams to participate successfully in various CTF events throughout the year.

During the panel, the MCC students fielded questions from the audience about degree programs at George �鶹��Ƶ, the benefits of technical certifications, and how to best prepare for a career in cybersecurity. “Most cyber jobs are here in the D.C. area,” noted Knoff, pointing out one benefit of pursuing cybersecurity at George �鶹��Ƶ.

CyberSlam fosters the cultivation of cybersecurity talent in Northern Virginia, where the industry is booming. This year’s event particularly underscored the opportunities in both public and private sectors that are preparing the next generation of cybersecurity professionals, ensuring they are equipped with the skills and knowledge needed to thrive in this rapidly growing field.

“I would like to thank CCI NoVA Node, specifically Liza Durant and Jordan �鶹��Ƶ for sponsoring the amazing raffle prizes and their support for the Virginia CyberSlam, and also George �鶹��Ƶ’s Elizabeth Dean for making this event possible,” said Marden. “We could not have done this with the support of George �鶹��Ƶ, CCI, our presenters, sponsors, and of course our wonderful Virginia cybersecurity educators and students.”

Explore Cybersecurity at George �鶹��Ƶ

Topics

Cybersecurity

Department of Cyber Security Engineering

Commonwealth Cyber Initiative (CCI)

Students

Chef’s KISS: Statistics PhD candidate creates new recipe for complex spatial data analysis

Teresa Donnellan — Thu, 06 Mar 2025 20:40:50 +0000

Chef’s KISS: Statistics PhD candidate creates new recipe for complex spatial data analysis Teresa Donnellan Thu, 03/06/2025 - 15:40

In This Story

Ben Seiyon Lee

Jin Hyung Lee, a PhD candidate in the Department of Statistics, in collaboration with his advisor, Assistant Professor Ben Seiyon Lee, has developed novel statistical methodology for analyzing large-scale complex data that can be up to 3,600 times faster than the current standard. He will receive the Korean International Statistical Society (KISS) 2024 Outstanding Student Paper Award for his work and present about it at the American Statistical Association’s Joint Statistical Meetings (JSM) 2025 in Nashville, Tennessee.

The primary application of Lee’s methodology is in the analysis of high-resolution satellite image data, which is inherently complex and high-dimensional. A major challenge lies in predicting values at unobserved locations within the satellite imagery. While traditional spatial modeling methods are computationally intensive and time-consuming, Lee's novel approach offers a solution that is significantly faster. This development has significant implications for various fields, such as atmospheric sciences, ecology, public health, and medical imaging.

Jin Hyung Lee, photo provided

Lee’s research focuses on the application of variational inference to massive spatial datasets using a unique machine learning algorithm. Variational inference is a method used to approximate complex probability distributions when true inference is not possible, making it a powerful method for modeling complex spatial datasets. The proposed approach significantly improves the computational efficiency of spatial data analysis, achieving speeds up to three orders of magnitude faster than existing methods while maintaining comparable accuracy.

Lee's method can be used to predict weather patterns in remote and rapidly changing environments like the Gulf Coast, southern California, and polar regions like Alaska or Antarctica. The ability to process high-dimensional data quickly and accurately is crucial for making timely and informed decisions in these regions, especially when they are at risk of geophysical hazards such as hurricanes, wildfires, or sea ice loss. Additionally, the methodology has potential applications in infectious disease modeling and public health data analysis, where rapid and accurate predictions are essential for effective intervention and policymaking.

Lee's research is part of his broader dissertation work, which focuses on developing and applying variational inference techniques to various statistical models. His goal is to simplify these methods, making them more accessible to non-experts and applicable to a wide range of data types. By doing so, Lee aims to bridge the gap between advanced statistical methodologies and practical applications, enabling more efficient and effective data analysis across different fields.

At the JSM 2025, Lee will have the opportunity to present his research to a large audience of statisticians from around the world. The conference, which attracts over 5,000 participants annually, provides a platform for sharing cutting-edge research, networking with peers, and exploring new developments in the field of statistics. Lee’s presentation will be part of a session hosted by KISS, where he and other award recipients will showcase their work and receive their awards.

Receiving the KISS Outstanding Student Paper Award is a significant milestone in Lee’s academic journey. It not only recognizes his contributions to the field but also provides him with valuable exposure and opportunities for future collaboration. As Lee continues to advance his research, his innovative methodologies and their applications are poised to make a lasting impact on the field of statistics and beyond.

Topics

Department of Statistics

Computational statistics

International Space Station experiment may improve climate science predictions

Teresa Donnellan — Wed, 05 Mar 2025 20:33:24 +0000

International Space Station experiment may improve climate science predictions Teresa Donnellan Wed, 03/05/2025 - 15:33

The earth’s atmosphere is full of aerosols, both from natural events such as volcanic eruptions and human activity such as burning fossil fuels. The Department of Mechanical Engineering’s Jeffrey Moran is planning an experiment on the International Space Station (ISS) to better understand what factors drive aerosols from place to place, which could enable climate scientists to predict the overall effect of aerosols on climate with less uncertainty.

“Aerosols can exacerbate or alleviate climate change, depending on what they’re made of,” said Moran. “For example, carbon soot aerosols can intensify warming by absorbing sunlight that would otherwise be reflected back into space. Other aerosols, such as sulfates, can reduce warming by increasing solar reflection.”

Aerosols’ location also matters—for example, many aerosols can act as nuclei for clouds to form, and this can exert an overall warming or cooling effect in different locations. “The overall effect aerosols will have on the climate is hard to predict, partly because we don’t have a full understanding of what factors cause them to migrate from one place to another,” said Moran.

Jeff Moran. Photo by Evan Cantwell/Office of University Branding

The experiment is specifically examining whether and how temperature gradients influence the movement of certain aerosols through the atmosphere. Moran’s team, along with his collaborator, David Warsinger of Purdue University, and a team of undergraduate students there, are designing an apparatus inside of which aerosols of a certain size can be observed across a temperature gradient. They recently completed a fit check to confirm that the apparatus both fits in the ISS microscope’s enclosure and that the aerosol container, or cuvette, is transparent enough for observation.

Moran said the experiment could help climate researchers understand how temperature differences, which are common in the atmosphere, cause particles to move through the air.

“If the effect of the temperature gradient is significant, it could give us a clearer indication of how aerosols will affect the climate in the future. It could also inform geoengineering proposals, which would seek to mitigate climate change by intentionally injecting aerosols into the atmosphere, since we would have a better sense of where they will go, and thus what their overall effect will be,” said Moran. “It could also offer a way to control aerosols on the ground, for example, by using thermophoresis to collect virus-laden aerosols from the HVAC systems of hospitals.”

The team’s apparatus includes a 3D-printed stage with hot and cold fixtures on either end. A prototype cuvette is mounted between them in the enclosure of the microscope. Crucially, the apparatus fits snugly inside the microscope enclosure, which is necessary for the experiment to be carried out in microgravity.

Crystals of the NaCl sample (see Figure 2), imaged using the 2x objective on the Keyence BZ-X810 microscope (nearly identical to the one on the ISS).

“The capstone team did a wonderful job,” Moran added.

The team tested the cuvette’s transparency by placing sodium chloride, table salt, inside it and taking images with the microscope. To verify airtightness, the team placed the salt-filled cuvette in a beaker of water and observed that the salt (which is soluble in water) remained in its solid form, indicating no water had leaked into the cuvette.

“We obviously don’t want to pose an inhalation hazard to the ISS crew, so it’s important that the particle samples be packaged on the ground in containers that are as close to airtight as possible,” said Moran.

“These images are a promising first step, as they demonstrate that aerosols in the range of sizes we are considering (10-100 µm) can be clearly imaged through the prototype cuvettes using a microscope nearly identical to the one currently on the ISS,” said Moran.

From here, Moran, Warsinger, and a new team of students will work to improve the cuvette fabrication and polishing procedures. They will also test alternative cuvette materials such as quartz, which may be thinner and more transparent, thereby allowing them to take clearer images at a higher magnification. This would enable them to examine even smaller aerosol particles. The team is also refining the design of the cuvettes to improve optical transparency and minimize the thickness of the windows.

“We want to make the cuvettes smaller, bringing the hot and cold surfaces as close together as we can get them, because it's not really clear how steep of a temperature gradient we're going to need to induce motion,” Moran explained.

In addition, the team is finalizing their choice of aerosol materials to send to the ISS. They are currently set on using carbon soot, sodium chloride, and silica due to their relevance in atmospheric aerosols and climate science.

“Carbon soot is one of the most important materials in terms of climate, because it is known to intensify warming,” Moran said, adding that sodium chloride enters the atmosphere as sea spray and can drift to faraway locations, such as the North Pole, intensifying the warming that is already happening at a rapid pace there. Silica is a common ingredient in sand and thus a common material in the atmosphere as well. Overall, the team aims to strike a balance between having materials that can be studied safely on the ISS while also yielding insights into aerosol dynamics in the atmosphere.

The College of Engineering and Computing will cover Moran’s progress leading up to the experiment in space, scheduled for fall 2025 or spring 2026.

In This Story

Jeffrey Moran

Topics

mechanical engineering

Keeping data safe by keeping it separate

Teresa Donnellan — Tue, 25 Feb 2025 15:35:36 +0000

Keeping data safe by keeping it separate Teresa Donnellan Tue, 02/25/2025 - 10:35

Developing a robust cybersecurity system demands a lot of data, but keeping all that data in a single location introduces the risk of a data breach. Luckily, �鶹��Ƶ researcher Zhuangdi Zhu in the Department of Cyber Security Engineering specializes in federated learning, a method of decentralized machine learning that allows multiple entities to collaboratively train a machine learning model without sharing their raw data.

Zhu is using this expertise to assist Wajih Ul Hassan of the University of Virginia develop scalable host-based intrusion detection systems (HIDS). Together, Zhu and Hassan are leveraging artificial intelligence (AI) to enhance a given host’s cybersecurity while ensuring data privacy.

HIDS monitor and analyze the activities within a single host or device. These systems use data, such as system logs, web browsing logs, and process logs, to detect irregularities that may indicate a security threat. For example, Zhu noted, a HIDS might detect malicious software that prompts a browser to download a harmful PDF file. By identifying such irregularities, the HIDS can prevent potential security breaches. Central management of these logs through managed security service providers (MSSPs) on cloud servers, however, could potentially breach user privacy. Enter: federated learning.

Zhuangdi Zhu, photo provided

In a federated learning system, each entity trains a local model on its own data and only shares the model updates with a central server. Local models are trained on individual datasets, and their findings are sent to a central server, Zhu explained. The server then aggregates these parameters to create a global model, which is subsequently sent back to the local models. This iterative process continues until a robust and accurate model is developed. Federated learning thus ensures sensitive data remains within the local models, addressing privacy concerns and complying with data regulations.

The project, funded by the Commonwealth Cyber Initiative, is set to span one year, starting in the summer of 2025. In addition to developing a federated learning framework for HIDS, the team plans to create an open-source dataset for intrusion detection, which will serve as a valuable resource for future research and development in this field.

Zhu's work on decentralized machine learning not only addresses critical privacy concerns but also paves the way for more secure and efficient threat detection systems. Her work on this project represents a significant step forward in the ongoing effort to protect digital environments from evolving cyber threats.

In This Story

Zhuangdi Zhu

Learn more about George �鶹��Ƶ's Department of Cyber Security Engineering

Topics

Cybersecurity

Commonwealth Cyber Initiative (CCI)

Machine Learning

Artificial Intelligence

Research

Using AI to uncover human smuggling networks

Teresa Donnellan — Wed, 15 Jan 2025 22:39:59 +0000

Using AI to uncover human smuggling networks Teresa Donnellan Wed, 01/15/2025 - 17:39

As vulnerable members of society, migrants are subject to exploitation. Professor Carlotta Domeniconi in the Department of Computer Science at �鶹��Ƶ is using artificial intelligence (AI) to combat that abuse. She is the principal investigator (PI) on a groundbreaking project aimed at understanding and modeling human smuggling networks.

Carlotta Domeniconi

In collaboration with the Department of Homeland Security’s Criminal Investigations and Network Analysis (CINA) Center for Excellence, Domeniconi and co-PIs Guadalupe Correa-Cabrera and Sean Luke are developing advanced machine learning techniques to analyze publicly available data and uncover the intricate workings of human smuggling networks between Mexico and the United States. While Domeniconi and Luke, both computer science faculty, bring expertise in machine learning, Correa-Cabrera is a professor in George �鶹��Ƶ's Schar School of Policy and Government specializing in border studies, U.S.-Mexico relations, international security, migration studies, and illicit networks.

A significant challenge in this project is the sparsity and complexity of the data, said Domeniconi. Criminal networks are large, dynamic, and constantly evolving.

"The very nature of criminal networks—their secret nature—makes it practically impossible to gather useful data," she explained. This challenge led to the team’s innovative approach of using publicly available text data—past cases of prosecuted smugglers at both the state and federal levels in the United States as well as social media posts and news articles by and about smugglers. They are collecting data from the 1980s to 2024 to analyze the temporal dynamics of these networks.

“There are videos of smugglers actually showcasing people traveling across very difficult areas from Central America,” Domeniconi noted as an example. “You can extract a lot of information about the routes they take, where they cross the borders, and the means they use."

The team is using natural language processing (NLP) and deep learning models that are pre-trained on vast amounts of text data, enabling them to understand and extract meaningful information from the data they are collecting. Two doctoral students will support the co-PIs, one from computer science and the other from the Schar School. The team is working on automating the process of building and analyzing these criminal networks using large language models and graph mining techniques. By visualizing these networks as knowledge graphs, the team aims to mine behavioral patterns and understand how these networks adapt to changes in policy enforcement. Ultimately, this project could provide a methodology applicable to various criminal networks beyond human smuggling.

Guadalupe Correa-Cabrera

“The current project bridges the gap between machine learning and social science,” explained Correa-Cabrera. “We can replicate this same model and methodology to analyze legal cases of drug smuggling, money laundering, and arms trafficking, thereby modeling and explaining the evolution of these activities over the past few decades.”

Domeniconi's pioneering work in modeling human smuggling networks is a testament to the power of AI in addressing complex social issues. By merging domain expertise with state-of-the-art machine learning techniques, this project not only promises to unravel the clandestine operations of human smuggling but also to create a robust methodology applicable to various criminal networks. The insights gained from this research could significantly enhance national security measures, offering a strategic advantage in the fight against human smuggling and those who perpetrate this crime.

Start your George �鶹��Ƶ journey

In This Story

Carlotta Domeniconi

Sean Luke

Guadalupe Correa-Cabrera

Aisha Behr leverages dual roles in industry and academia to advance deep learning

Teresa Donnellan — Wed, 08 Jan 2025 15:17:30 +0000

Aisha Behr leverages dual roles in industry and academia to advance deep learning Teresa Donnellan Wed, 01/08/2025 - 10:17

With degrees from �鶹��Ƶ in civil engineering, systems engineering, and earth systems and geoinformation sciences, alumna Aisha Behr’s path to becoming a leader in artificial intelligence (AI) and machine learning is as diverse as it is inspiring. Now a director for the AI and machine learning business units at CGI, Behr also serves as an adjunct professor in George �鶹��Ƶ’s Department of Information Sciences and Technology (IST), teaching applied information technology (AIT) classes in machine learning and deep learning.

Behr started her career in government contracting, where she first encountered the burgeoning field of data science.

Her work required her to apply data science techniques to geospatial analysis, sparking a deep interest in the subject. While working full-time, Behr pursued a PhD in earth systems and geoinformation sciences at George �鶹��Ƶ. Her dissertation, which aimed to improve the SVD algorithm with spatial statistics, highlighted her innovative approach to integrating machine learning with spatial data.

Aisha Behr. Photo provided

While balancing her professional career with her academic pursuits, Behr found a new calling in teaching, starting as an adjunct after completing her PhD in 2020. Her industry experience proves invaluable in the classroom, allowing her to offer students a unique perspective on the practical applications of their studies.

"I wanted to make sure that students had a really good process and experience through the whole course," said Behr. By providing code templates and practical assignments, she ensures that students can confidently apply what they learn in the AI and machine learning field.

Behr's teaching philosophy is centered around making complex concepts accessible and engaging. She developed the courses AIT 636 Interpretable Machine Learning and AIT 746 Applied Deep Learning to emphasize hands-on learning.

“Dr. Behr has been an exemplary adjunct in the IST department, playing a critical role in developing our new applied AI technologies concentration for the master’s in AIT program, as well as several graduate courses,” said Ioulia Rytikova, an IST professor and the director of the master’s program in AIT.

Behr is excited about the potential of generative AI and other emerging technologies. She will continue integrating such advancements into her teaching, possibly through guest lectures and hands-on labs.

"I want our students to be able to have exposure to these very, very new technologies," she emphasized, noting that she likes to demonstrate to students how keeping up with technology can give them a competitive edge in their careers.

"I enjoy learning new concepts,” said Behr. “Being able to show a student that you can learn and then accelerate your career is why I like it so much."

Behr’s journey is a powerful example of how passion, adaptability, and a commitment to education can lead to a fulfilling and impactful career. Her dedication to bridging industry and academia ensures that her students are well-prepared to navigate and excel in the ever-evolving field of AI and machine learning.

Learn more about Information Sciences and Technology at George �鶹��Ƶ

In This Story

Ioulia Rytikova

Topics

information sciences and technology

Artificial Intelligence

applied information technology graduate programs

Machine Learning

Adjunct Faculty

Alumni

Teresa M. Donnellan

Brainpower inspires energy-efficient AI

In This Story

Related Stories

Topics

George �鶹��Ƶ’s Imaging and Vision Lab illuminates the future of digital production

In This Story

Topics

Innovative approaches propel STAT 250 to new heights at �鶹��Ƶ

In This Story

Topics

Cyber sweep: George �鶹��Ƶ students take top three spots in national transportation cybersecurity competition

In This Story

Related Stories

Topics

Nearly 500 high schoolers get expert advice at Virginia CyberSlam 2025

Related Stories

Topics

Chef’s KISS: Statistics PhD candidate creates new recipe for complex spatial data analysis

In This Story

Topics

International Space Station experiment may improve climate science predictions

In This Story

Related Stories

Topics

Keeping data safe by keeping it separate

In This Story

Related Stories

Topics

Using AI to uncover human smuggling networks

In This Story

Related News

Aisha Behr leverages dual roles in industry and academia to advance deep learning

In This Story

Related News

Topics