Georgia Tech researchers are making AI easier to understand through their work on Transformer Explainer. The free, online tool shows non-experts how ChatGPT, Claude, and other large language models (LLMs) process language. 

Transformer Explainer is easy to use and runs on any web browser. It quickly went viral after its debut, reaching 150,000 users in its first three months. More than 563,000 people worldwide have used the tool so far.

Global interest in Transformer Explainer continues when the team presents the tool at the 2026 Conference on Human Factors in Computing Systems (CHI 2026). CHI, the world’s most prestigious conference on human-computer interaction, will take place in Barcelona, April 13-17.

[Related: GT @ CHI 2026]

“There are moments when LLMs can seem almost like a person with their own will and personality, and that misperception has real consequences. For example, there have been cases where teenagers have made poor decisions based on conversations with LLMs,” said Ph.D. student Aeree Cho.

“Understanding that an LLM is fundamentally a model that predicts the probability distribution of the next token helps users avoid taking its outputs as absolute. What you put in shapes what comes out, and that understanding helps people engage with AI more carefully and critically.”

A transformer is a neural network architecture that changes data input sequence into an output. Text, audio, and images are forms of processed data, which is why transformers are common in generative AI models. They do this by learning context and tracking mathematical relationships between sequence components.

Transformer Explainer demystifies how transformers work. The platform uses visualization and interaction to show, step by step, how text flows through a model and produces predictions.

Using this approach, Transformer Explainer impacts the AI landscape in four main ways:

• It counters hype and misconceptions surrounding AI by showing how transformers work.
• It improves AI literacy among users by removing technical barriers and lowering the entry for learning about AI.
• It expands AI education by helping instructors teach AI mechanisms without extensive setup or computing resources.
• It influences future development of AI tools and educational techniques by providing a blueprint for interpretable AI systems.

“When I first learned about transformers, I felt overwhelmed. A transformer model has many parts, each with its own complex math. Existing resources typically present all this information at once, making it difficult to see how everything fits together,” said Grace Kim, a dual B.S./M.S. computer science student. 

“By leveraging interactive visualization, we use levels of abstraction to first show the big picture of the entire model. Then users click into individual parts to reveal the underlying details and math. This way, Transformer Explainer makes learning far less intimidating.”

Many users don’t know what transformers are or how they work. The Georgia Tech team found that people often misunderstand AI. Some label AI with human-like characteristics, such as creativity. Others even describe it as working like magic.

Furthermore, barriers make it hard for students interested in transformers to start learning. Tutorials tend to be too technical and overwhelm beginners with math and code. While visualization tools exist, these often target more advanced AI experts.

Transformer Explainer overcomes these obstacles through its interactive, user-focused platform. It runs a familiar GPT model directly in any web browser, requiring no installation or special hardware. 

Users can enter their own text and watch the model predict the next word in real time. Sankey-style diagrams show how information moves through embeddings, attention heads, and transformer blocks.

The platform also lets users switch between high-level concepts and detailed math. By adjusting temperature settings, users can see how randomness affects predictions. This reveals how probabilities drive AI outputs, rather than creativity.

“Millions of people around the world interact with transformer-driven AI. We believe that it is crucial to bridge the gap between day-to-day user experience and the models' technical reality, ensuring these tools are not misinterpreted as human-like or seen as sentient,” said Ph.D. student Alex Karpekov. 

“Explaining the architecture helps users recognize that language generated by models is a product of computation, leading to a more grounded engagement with the technology.” 

Cho, Karpekov, and Kim led the development of Transformer Explainer. Ph.D. students Alec Helbling, Seongmin Lee, Ben Hoover, and alumnus Zijie (Jay) Wang assisted on the project. 

Professor Polo Chau supervised the group and their work. His lab focuses on data science, human-centered AI, and visualization for social good.

Acceptance at CHI 2026 stems from the team winning the best poster award at the 2024 IEEE Visualization Conference. This recognition from one of the top venues in visualization research highlights Transformer Explainer’s effectiveness in teaching how transformers work.

“Transformer Explainer has reached over half a million learners worldwide,” said Chau, a faculty member in the School of Computational Science and Engineering. 

“I'm thrilled to see it extend Georgia Tech's mission of expanding access to higher education, now to anyone with a web browser.”

Published today in the journal Nature Communications, the paper “Trivalent Titanium in High-Titanium Lunar Ilmenite” confirms titanium in a reduced, trivalent state in a black, metal-rich lunar mineral called ilmenite. It’s a state only possible in low-oxygen environments, conditions researchers refer to as “reducing.”

“Models have suggested that these reducing conditions may have varied at different locations and times across the surface of the Moon,” says lead author Advik Vira, a graduate student in the School of Physics who recently earned his doctoral degree. “We hope our microscopy technique can be a valuable step in mapping and understanding the Moon’s 4.5-billion-year history.”

The team anticipates that their technique could be used on many of the lunar samples collected more than 50 years ago by the Apollo missions in addition to the Apollo Next Generation Samples — a group of lunar samples that have been stored under pristine conditions — and new samples from the planned Artemis missions, with Artemis II slated for launch this spring. The technique might also be applicable to samples collected from the far side of the Moon and returned in 2024 by the Chang’e-6 mission.

“The Moon holds clues not only to its own past, but also to the earliest eras of Earth’s evolution — history that has long since been erased from our planet,” Vira says. “This study is a step toward understanding the history of both and a reminder that there is still so much left to learn from the lunar rocks we’ve brought back to Earth.”

The School of Physics research team included corresponding authors Vira and Professor Phillip First; in addition to graduate student Roshan Trivedi; undergraduate students Gabriella Dotson, Keyes Eames, Dean Kim, and Emma Livernois; and Professor Zhigang Jiang, along with Institute for Matter and Systems Materials Characterization Facility Senior Research Scientist Mengkun Tian; School of Chemistry and Biochemistry Senior Research Scientist Brant Jones and Thom Orlando, Regents' Professor in the School of Chemistry and Biochemistry with a joint appointment in the School of Physics. 

The Georgia Tech team was joined by Addis Energy Senior Geochemist Katherine Burgess; Macalester College Assistant Professor of Geology Emily First; along with Lawrence Berkeley National Laboratory Research Scientist Harrison Lisabeth, Senior Scientist Nobumichi Tamura, and Postdoctoral Fellow Tyler Farr, who recently earned a Ph.D. from Georgia Tech’s George W. Woodruff School of Mechanical Engineering.

CLEVER research

The investigation began with a dark gray rock called a lunar basalt. Formed when ancient magma erupted on the Moon’s surface, minerals crystallized as it cooled — preserving key information in their structures. Billions of years later, the rock was brought to Earth by the 1972 Apollo 17 mission, where a small piece is now stored at Georgia Tech’s Center for Lunar Environment and Volatile Exploration Research (CLEVER), a NASA Solar System Exploration Research Virtual Institute (SSERVI) center led by Orlando.

As a NASA virtual institute, CLEVER supports researchers exploring lunar conditions and developing tools for the upcoming crewed Artemis missions, and provided the lunar samples for this research. The SSERVI also plays a critical role in training the next generation of planetary researchers: both Vira and Farr earned their Ph.D.s while on the CLEVER team.

“At CLEVER, we are very interested in understanding the impacts of space weathering,” Vira says. “We implemented modern sample preparation and advanced microscopy techniques to image samples at the atomic level, and were curious to apply it more broadly to the collection of Apollo rocks in the Orlando Lab. This sample caught our attention.”

“When we imaged an ilmenite crystal from the lunar basalt, what struck us first was how uniform and perfect the crystal structure was,” he recalls. “We found no defects from space weathering and instead saw an undamaged, pristine crystal — undisturbed for 3.8 billion years.”

To investigate further, the team analyzed small chips of the rock with Burgess, a member of the RISE2 SSERVI team and then a geologist at the U.S. Naval Research Laboratory. Using state-of-the-art electron microscopy and spectroscopy techniques, Vira determined the oxidation state of the elements in the ilmenite present. 

In spectroscopy measurements, each element leaves a distinct ‘signature,’ Vira explains. “When we brought our results back to Georgia Tech’s Materials Characterization Facility, Mengkun (Tian) noticed something unusual: the signature showed titanium might be present in the trivalent state.”

The presence of trivalent titanium had long been suspected in this lunar mineral. The team was intrigued. 

A new window into old rocks

With funding from Georgia Tech’s Center for Space Technology and Research (CSTAR), Vira returned to the U.S. Naval Research Laboratory to analyze additional samples. The results confirmed that more titanium was present than the mineral’s formula (FeTiO₃) predicts — indicating a portion of the titanium present was trivalent.

“That led me to place our measurements in terms of the broader geological context,” Vira shares. Working with First, Vira explored how ilmenite with trivalent titanium could help reconstruct the nature of ancient magmas from the Moon, especially the chemical availability of oxygen.

“Because its location on the Moon was noted during the Apollo mission, we know exactly where this rock is from, and we can determine how old the rock is,” he explains. “When coupled with our trivalent titanium measurements, we can use that information to estimate the reducing conditions for this specific region at the specific time our rock formed.”

If the upcoming Artemis missions return samples suitable for the team’s technique, these rocks could provide a new window into ancient lunar geology. The research also highlights that many lunar samples already on Earth could be reexamined to look for trivalent titanium.

“There is still so much to learn from the lunar samples we have already brought to Earth,” Vira says. “It’s a testament to the long-term value of each sample return mission. As technology continues to advance, this type of work will continue to give us critical insights into our planet and our place in the universe for years to come.”

DOI: 10.1038/s41467-026-69770-w

Funding: This work was directly supported by the NASA SSERVI under CLEVER. Researchers were also supported by the NASA RISE2 SSERVI and the Heising-Simons Foundation. Funding for collaborations between the U.S. Naval Research Laboratory and Georgia Tech for the investigation of lunar minerals was provided by the Georgia Tech Center for Space Technology and Research. Sample preparation was performed at the Georgia Tech Institute for Matter and Systems, which is supported by the National Science Foundation. This work utilized the resources of the Advanced Light Source, a user facility supported by the U.S. Department of Energy, Office of Science, Office of Basic Energy Sciences, and was supported in part by previous breakthroughs obtained through the Laboratory Direct.

The Georgia Scientific Computing Symposium (GSCS) held its annual meeting on February 21 in Athens. Since 2009, the event has hosted researchers from across the Peach State to showcase homegrown advances in scientific computing.

The symposium highlighted Georgia’s reputation as a computing innovation hub. People from around the world come to Georgia universities to lead computing research. By advancing science, engineering, medicine, and technology, their work improves communities at home and abroad.

Faculty and students from Georgia Tech, UGA, Georgia State University, and Emory University presented at the symposium. Georgia Tech participants came from the colleges of Computing, Engineering, and Sciences.

This year’s organizers agreed to meet in Atlanta for the 2027 symposium. Georgia Tech’s School of Computational Science and Engineering (CSE) will host the 19th GSCS.

“From healthcare to computer chip design, scientific computing underpins many of the technological advances we see in our lives,” said Professor Edmond Chow, associate chair of the School of CSE.

“Scientific computing provides the mathematical models, simulations, and data‑driven tools that make modern innovation possible. It allows people to analyze complex systems, test ideas virtually before building them, and make faster, more accurate decisions across nearly every sector of society.”

Professor Haomin Zhou and Assistant Professor Helen Xu delivered two of the symposium’s five plenary talks. 

Zhou presented a new method for solving the Schrödinger equation, a landmark equation in quantum mechanics. Drawing inspiration from the mathematics used in generative artificial intelligence models, his approach develops an algorithm that more effectively simulates waves, particle motion, and other physical systems.

Xu focused on improving how computers move and organize data during complex calculations. Her work uses “cache-friendly” layouts that help computers access data more efficiently, boosting performance for scientific and engineering applications.

“Speaking at GSCS was a great opportunity,” Xu said. “The symposium fostered connections within the scientific computing community and gave us a chance to share exciting research.”

The symposium showcased student work through a poster blitz and a poster session. During the blitz, 36 students each had one minute to introduce their research to the full audience. They then shared more details about their research during the poster session.

The student projects showed the range of fields supported by scientific computing. The session also provided attendees with an opportunity to connect and expand their professional networks, helping grow the field’s future impact.

“As an aerospace engineer by training and aspiring computational scientist, GSCS gave me the platform to network with other researchers in the field while showcasing my own research,” said M.S. student Kashvi Mundra. 

“I was able to connect with scientists across different disciplines whose work intersects with my own in unexpected ways. Those conversations pushed my thinking beyond my own lab's perspective, helping me see my work on physics-informed machine learning for inverse problems in a broader scientific computing context.”

Georgia Tech students who presented posters included:

Abir Haque (CSE), Massively Parallel Random Phase Approximation Correlation Energy via Lanczos Quadrature

Antonio Varagnolo (CSE), Physics-Enhanced Deep Surrogates for the Phonon Boltzmann Transport Equation

Ben Burns (CSE), Infinite-Dimensional Stein Variational Inference with Derivative-Informed Neural Operators

Ben Wilfong (CSE), Shocks without Shock Capturing; Compressible Flow at 1 quadrillion Degrees of Freedom without Loss of Accuracy

Daniel Vickers (CSE), Highly-Parallel Fluid-Solid Interactions for Compressible Flows

Eric Fowler (CSE), High-Performance Tensor Contractions in Computational Chemistry

Haoran Yan (Math), Understanding Denoising Autoencoders through the Manifold Hypothesis: A Geometric Perspective

Kashvi Mundra (CSE), Autoregressive Multifidelity Neural Surrogate Modeling under Scarce Data Regimes

Sebastián Gutiérrez Hernández (Math/CSE), PDPO: Parametric Density Path Optimization

Vivian Zhang (AE), Multifidelity Operator Inference: Non-Intrusive Reduced Order Modeling from Scarce Data

Xian Mae Hadia (CSE), Data Efficiency of Surrogate Models: Learning Physics Data from Full Field Data vs. Inductive Bias from Approximate PDE Solvers

Xiangming Huang (CSE), Neural Operator Accelerated Evolutionary Strategies for PDE-Constraint Optimization

Zhaiming Shen (Math), Understanding In-Context Learning on Structured Manifolds: Bridging Attention to Kernel Methods

Zhongjie Shi (Math), Towards Understanding Generalization in DP-GD: A Case Study in Training Two-Layer CNNs

New cybersecurity research from the Georgia Institute of Technology, however, revealed that crews aboard commercial vessels were often not adequately prepared to manage cyberattacks effectively due to systemic training gaps.

The findings are based on interviews conducted by researchers with more than 20 officer-level mariners to assess the maritime industry’s readiness to handle cybersecurity attacks at sea.

"Historically, cybersecurity research has focused heavily on cyber-physical systems like cars, factories, and industrial plants, but ships have largely been overlooked,” said Anna Raymaker, Ph.D. student and lead researcher.

“That gap is concerning when more than 90% of the world’s goods travel by sea. Recent incidents, from GPS spoofing to ships linked to subsea cable disruptions, show that maritime systems are increasingly part of the global cyber threat landscape.”

The researchers proposed four practical strategies to strengthen maritime cyber defenses and close the training gaps. Their findings were presented recently at the ACM SIGSAC Conference on Computer and Communications Security (CCS).

1. Make Cybersecurity Training Actually Maritime

Many of those interviewed for the study described current cybersecurity training as “boilerplate” — generic modules that don’t reflect real shipboard risks. 

Researchers recommend:

• Role-specific instruction: Navigation officers should learn to detect and identify GPS spoofing. Engineers should focus on vulnerabilities in remotely monitored systems.
• Bridging IT and Operational Technology: Crews need to understand how attacks on IT systems can trigger physical consequences in operational technology — including collisions, groundings, or explosions.
• Hands-on delivery: Replace passive PowerPoints with drills and in-person exercises that build muscle memory.
• Accessible standards: Training must account for the wide range of educational backgrounds across crews and be standardized across ranks.

2. Move Beyond “Call IT”

At sea, crews can’t simply escalate a cyber incident to a shore-based IT department and wait. Operational resilience requires onboard readiness.

Researchers recommend:

• Vessel-specific response plans: Ships need clear, actionable protocols for threats such as AIS jamming or radar manipulation.
• Military-style drills: Adopting MCON (Emission Control) exercises — used by the U.S. Military Sealift Command — can train crews to operate safely without electronic systems.
• Stronger connectivity controls: High-bandwidth satellite systems like Starlink introduce new risks. Clear policies and network segregation are essential to prevent new entry points for attackers.

Related Article: When GPS lies at sea: How electronic warfare is threatening ships and their crews by Anna Raymaker

3. Create Unified, Ship-Specific Regulations

Maritime cybersecurity regulations are often reactive and fragmented. Researchers argue the industry needs a cohesive, domain-specific framework.

Key recommendations include:

• A unified global model: Like the energy sector’s NERC CIP standards, a maritime framework could mandate baseline controls such as encryption, network segmentation, and anonymous incident reporting.
• Rules built for real crews: Regulations designed for large naval operations don’t translate well to smaller merchant or research vessels. Standards must reflect actual shipboard conditions.
• Future-proofing requirements: Autonomous ships and remotely operated vessels expand the cyber-physical attack surface. Regulations must proactively address these emerging technologies.

4. Invest in Maritime-Specific Cyber Research

Finally, the researchers stress that long-term resilience requires deeper technical research focused on maritime systems.

Priority areas include:

• Real-time intrusion detection systems tailored to shipboard protocols.
• Proactive security risk assessments of interconnected onboard systems.
• Cyber-physical modeling to better understand cascading failures in complex maritime environments.

The Bottom Line

Cyber threats at sea are no longer hypothetical. Mariners report real-world incidents ranging from GPS spoofing to ransomware that disrupts global trade.

“Through our interviews with mariners, I saw firsthand how much dedication and pride they take in their work,” said Raymaker. “Our goal is for this research to serve as a call to action for researchers, policymakers, and industry to invest more attention in maritime cybersecurity and support the people who risk their lives every day to keep global trade, food, and energy moving."

A Sea of Cyber Threats: Maritime Cybersecurity from the Perspective of Mariners was presented at CCS 2025. It was written by Raymaker and her colleagues, Ph.D. students Akshaya Kumar, Miuyin Yong Wong, and Ryan Pickren; Research Scientist Animesh Chhotaray, Associate Professor Frank Li, Associate Professor Saman Zonouz, and Georgia Tech Provost and Executive Vice President for Academic Affairs Raheem Beyah.

Amino acids also play a critical role commercially where they are manufactured and added to pharmaceuticals, dietary supplements, cosmetics, animal feeds, and industrial chemicals — an energy-intensive process leading to greenhouse gas emissions, resource consumption, and pollution.

A landmark new system developed at Georgia Tech could lead to an alternative: a commercially scalable, environmentally sustainable method for amino acid production that is carbon negative, using more carbon than it emits.

The breakthrough builds on a method that the team pioneered in 2024 and solves a key issue – increasing efficiency to an unprecedented 97% and reducing the bioprocess cost by over 40%. It’s the highest reported conversion of CO2 equivalents into amino acids using any synthetic biology system to date.

Published in the journal ACS Synthetic Biology, the study, “Cell-Free-Based Thermophilic Biocatalyst for the Synthesis of Amino Acids From One-Carbon Feedstocks,” was led by Bioengineering Ph.D. student Ray Westenberg and Professor Pamela Peralta-Yahya, who holds joint appointments in the School of Chemistry and Biochemistry and School of Chemical and Biomolecular Engineering. The team also included Shaafique Chowdhury (Ph.D. ChBE 25) and Kimberly Wennerholm (ChBE 23); alongside University of Washington collaborators Ryan Cardiff, then a Ph.D. student and now a Chain Reaction Innovations Fellow at Argonne National Laboratory, and Charles W. H. Matthaei Endowed Professor in Chemical Engineering James M. Carothers; in addition to Pacific Northwest National Laboratory Synthetic Biology Team Leader Alexander S. Beliaev.

"This work shifts the narrative from simply reducing carbon emissions to actually consuming them to create value,” says Peralta-Yahya. “We are taking low-cost carbon sources and building essential ingredients in a truly carbon-negative process that is efficient, effective, and scalable.”

Heat-Loving Organisms

The work builds on the cell-free technology the team used in their earlier study. “Previously, we discovered that a system that uses the machinery of cells, without using actual living cells, could be used to create amino acids from carbon dioxide,” Peralta-Yahya explains. “But to create a commercially viable system, we needed to increase the system’s efficiency and reduce the cost.”

The team discovered that bits of leftover cells were consuming starting materials, and — like a machine with unnecessary gears or parts — this limited the system’s efficiency. To optimize their “machine,” the team would need to remove the extra background machinery.

"Leftover cell parts were using key resources without helping produce the amino acids we were looking for,” says Peralta-Yahya. “We knew that heating the system could be one way to purify it because heat can denature these components.”

The challenge was in how to protect the essential system components from the high temperatures, she adds. “We wondered if introducing enzymes produced by a heat-loving bacterium, Moorella thermoacetica, might protect our system, while still allowing us to denature and remove that inefficient background machinery.”

The results were astounding: after introducing the enzymes, heating and “cleaning” the system, and letting it cool to room temperature, synthesis of the amino acids serine and glycine leaped to 97% yield — nearly three times that of the team’s previous system.

Scaling for Sustainability

To make the system viable for large-scale use, the team also needed to reduce costs. “One of the most costly components in this system is the cofactor tetrahydrofolate (THF),” Peralta-Yahya shares. “Reducing the amount of THF needed to start the process was one way to make the system more inexpensive and ultimately more commercially viable.”

By linking reaction steps so waste from one step fueled the next, the team devised a method to recycle THF within the system that reduces the amount of THF needed by five-fold — lowering bioprocessing costs by 42%.

“This decrease in cost and increase in yield is a critical step forward in creating a method with real potential for use in industry and manufacturing,” Peralta-Yahya says. “This system could pave the way for moving this carbon-negative technology out of the lab and onto the continuous, industrial scale."

Funding: The Advanced Research Project Agency-Energy (ARPA-E); U.S. Department of Energy; and the U.S. Department of Energy, Office of Science, Biological and Environmental Research Program.

DOI: https://doi.org/10.1021/acssynbio.5c00352

Set in the heart of Tech Square on the Georgia Tech campus, this year’s event explores how energy systems, materials, technologies, supply chains, and policy must evolve in response to AI’s accelerating impact. As digital infrastructure expands and computation intensifies, the need for reliable, resilient, and sustainable power has never been more urgent. 

“Energy Day reflects Georgia Tech’s strength in connecting world-class research in materials and components with the infrastructure and partnerships needed to translate discovery into scalable energy technologies that serve industry, society, and the future economy,” said Eric Vogel, executive director of the IMS and the Hightower Professor in Materials Science and Engineering. 

Energy Day 2026 also marks an important milestone with the introduction of its first group of corporate sponsors: GE Vernova, Southern Company, Georgia Power, ExxonMobil, Southwire Spark, Gems Setra, and Tektronix. Their support reflects a shared commitment to advancing energy solutions. 

“Tektronix is excited to be part of Energy Day because advancing the future of energy starts with precise measurement and trusted insights,” said Christopher Bohn, president of Tektronix. “From power electronics and high voltage systems to grid scale renewables and AI driven control technologies, the breakthroughs discussed here directly align with the innovations we support through our products and solutions. Collaborating with Georgia Tech allows us to engage early with emerging research and the next generation of engineers—critical collaborators in building a cleaner, smarter, and more resilient energy ecosystem.”

The keynote address will be delivered by Vanessa Z. Chan, a nationally recognized leader at the intersection of innovation, commercialization, and emerging technologies. Chan will provide insights on accelerating technological discovery, emphasizing how AI is transforming energy and materials design. She will discuss how commercialization strategies must rapidly evolve across multidisciplinary energy domains from grid modernization to advanced batteries and clean manufacturing.

Building on the themes introduced in the keynote, the program transitions into a fireside chat with Georgia Tech EVPR Tim Lieuwen featuring Amit Kulkarni and Jim Walsh. Kulkarni is vice president of Product Management and Strategy for the Gas Power business within GE Vernova, where he oversees the world’s largest portfolio of power generation equipment. Walsh, vice president of GE Vernova’s Consulting Services, leads teams providing innovative solutions across the full spectrum of power generation, delivery, and utilization.

Next comes a policy-focused panel that will explore the surge in power demand driven by AI, how the United States is addressing today’s most urgent energy challenges, and the long-term implications of today’s decisions for a sustainable energy future. Bringing together leading voices in U.S. environmental and energy policy, the panel features Joe Aldy of Harvard University and former special assistant to the president for Energy and Environment; Al McGartland of New York University’s Institute for Policy Integrity and former Environmental Protection Agency lead economist and director of the National Center for Environmental Economics; and Kevin Rennert, fellow and director of the Comprehensive Climate Strategies Program at Resources for the Future and former staff member on the U.S. Senate Committee on Energy and Natural Resources.

The second panel focuses on critical materials — the foundation of advanced energy systems and digital technologies. As AI, data centers, and advanced energy technologies drive demand for critical materials, securing them now requires integration and coordination across the entire value chain. Panelists include Rachel Galloway, British consul general in Atlanta; Vijay Murugesan, head of Materials Intelligence and Digital Innovation at Amazon; Colin Spellmeyer, executive strategic sourcing leader at GE Vernova;  Charles Sims, Tennessee Valley Authority Distinguished Professor of Energy and Environmental Policy at the University of Tennessee; and Nortey Yeboah, principal engineer at Southern Company. Together, they will offer perspectives on the policy and economic frameworks shaping the energy supply chain, from developing raw resources to manufacturing the technologies essential to future energy systems.

In the afternoon, participants can dive deeper into specialized topics through three focused technical tracks. 

• “Meeting the Demand for Power” will examine how emerging technologies, advanced nuclear systems, and renewable integration can work together to deliver reliable, resilient electricity.
• “Data Center Infrastructure and Resources” will explore innovations in thermal management technologies, energy-efficient computing, and the broader resource impacts of expanding digital infrastructure.
• “Grid Technologies and Markets” will highlight strategies for strengthening grid capacity, incorporating demand-side management, and optimizing carbon performance as energy systems evolve.

“Meeting the rapidly rising electricity demand driven by AI requires bold ideas, coordinated action, and research that moves at the speed of innovation,” said Yuanzhi Tang, executive director of the SEI. “Energy Day 2026 brings together the people and expertise needed to shape resilient, sustainable energy systems for the future. At Georgia Tech, we see this event as a catalyst for new partnerships, new solutions, and a shared commitment to strengthening the nation’s energy foundation.”

Energy Day 2026 is designed for researchers advancing emerging energy technologies, policymakers navigating shifting regulatory and geopolitical landscapes, industry professionals seeking insight into emerging tools and supply chains, and students preparing to enter one of the most consequential sectors of the decade. It also welcomes anyone interested in AI, sustainability, electrification, and critical materials. 

Join us to explore the future of energy. To learn more and register, visit: Energy Day 2026.

In the International Disaster Reconnaissance (IDR) course, students now use Filio, a platform built by School of Computing Instruction Senior Lecturer Max Mahdi Roozbahani, to capture immersive 360° media, photos, and video that transform real disaster sites in India and Nepal into living digital classrooms. 

Offered by the School of Civil and Environmental Engineering and taught by IDR director and Regents’ Professor David Frost, the course pairs traditional fieldwork with Roozbahani’s expertise in immersive technology and data-driven learning, transforming on-the-ground observations into reusable, interactive educational resources. 

How Computing Can Capture Data 

Disasters are not only physical events; they are also information events, Roozbahani says. Effective response and long-term resilience depend on the ability to observe, record, and communicate critical data under pressure. Georgia Tech’s IDR course pairs structured on-campus preparation with international field experiences, enabling students to study the cascading effects of major disasters, including how local building practices, governance, and culture shape damage and recovery. 

“When students step into a disaster zone, they learn quickly that resilience is a systems problem: physical, social, and informational. Our job in computing is to help them capture and reason about that system responsibly,” Roozbahani said. 

Learning from the 2025 Himalayas Expedition 

During spring break last year, the cohort traveled along the Teesta River corridor in Sikkim, India. The region is shaped by steep terrain, fast-moving water, and critical infrastructure in narrow valleys. 

The visit followed the October 2023 glacial lake outburst flood from South Lhonak Lake, which destroyed the Teesta III hydropower dam and impacted downstream towns, including Dikchu and Rangpo. Field stops across India included Lachung, Chungthang, Dikchu, Rangpo, Gangtok, and New Delhi. 

Students explored both upstream and downstream consequences. 

Upstream, the team examined how steep terrain and river confinement amplify flood forces, creating cascading risks for infrastructure. Using Filio’s interactive 360° media, students captured conditions in Lachung and Chungthang, allowing viewers to explore the landscape through a 360° photo and 360° video that reveal how topography and river dynamics intensify disaster impacts. 

They studied community-scale effects downstream, including damaged buildings, disrupted access, and prolonged recovery timelines. 

Rangpo offered a glimpse of recovery in motion, with materials staged for rebuilding bridges and roads essential to commerce and emergency response.

Ryan Punamiya (CS 2025) and Summer Abramson, a third-year computational media student, have been honored by the Computing Research Association (CRA) through its 2025–2026 Outstanding Undergraduate Researcher Award (URA) program. 

Punamiya was named a runner-up for the prestigious award, while Abramson received an honorable mention among hundreds of applicants from universities across North America. 

The CRA Outstanding Undergraduate Researcher Award program recognized eight awardees in 2026, along with eight runners-up, nine finalists, and over 200 honorable mentions from thousands of applications.  

Advancing Robotics Research 

Punamiya knew early on that he didn’t want to wait until starting his Ph.D. to do meaningful and impactful robotics research.  

Punamiya joined the Robot Learning and Reasoning Lab (RL2) directed by Assistant Professor Danfei Xu. While there, he contributed to the lab’s Meta-sponsored EgoMimic project, which trains robots to perform human tasks using recordings captured by Meta’s Project Aria research glasses. 

Punamiya is also the first author of a paper accepted to the 2025 Conference on Neural Information Processing Systems (NeurIPS), one of the world’s most prestigious artificial intelligence (AI) and machine learning conferences. 

“Ryan is the strongest undergraduate I've worked with,” Xu said, “including students who went on to Stanford, Berkeley, and leadership roles in major tech companies. He’s already operating at the level of a strong third-year Ph.D. student.” 

Punamiya said it was a challenge to balance his undergraduate coursework with his research in Xu’s lab. 

“You get out how much you put in,” he said. “I built my class schedule to give myself as much time to do research as possible. It also boils down to having the right research mentors. 

“(Xu) never saw me as an undergrad who’s just there to do grunt work. I was fortunate he saw my curiosity and cultivated me as a researcher. That’s really how you get more undergrads motivated to research — giving them the chance to be independent and explore ideas of their own.” 

Punamiya said his work in Xu’s lab has already helped him identify the research areas he wants to focus on as he considers his next steps. He will continue developing generalized training models for robots using human data so they can perform tasks instantly upon deployment. 

"The amount of data needed to train a robot is difficult to obtain even for top industry companies," he said. "We have embodied robot data available in billions of humans. With the advent of extended reality devices, we can get a scalable source of diverse interactions within environments."

Punamiya graduated in December and recently started an internship at Nvidia. He mentioned he has been accepted into several Ph.D. programs, including Georgia Tech, and he is choosing where to continue his research. 

“It’s the first time my research has been acknowledged externally by the robotics community,” he said. “It’s good to know the problem I’m working on is important, and that motivates me. Robotics is an exciting field. We are doing things now that two years ago were difficult to do.” 

Researching Inclusion in Computing Education 

Abramson conducts research in the People-Agents Research for Computing Education (PARCE) Laboratory under the mentorship of Pedro Guillermo Feijóo-García, a faculty member in the School of Computing Instruction. He and the Associate Dean for Undergraduate Education, Olufisayo Omojokun, nominated her for the award. 

Her work focuses on the intersection of computing education and human-AI interaction, where she’s been exploring ways to create more equitable technology. 

“This is such a huge milestone, and I couldn't be prouder of Summer,” Feijóo-García said. “Mentoring her for almost two years has been an amazing experience.” 

Abramson has received the Georgia Tech President’s Undergraduate Research Award (PURA) twice, which supports her research exploring how user-centered design curricula can help address attrition among women in computing.

“I’ve had the amazing opportunity to pursue research at the intersection of student identity, community belonging, and how we can build tools that support our diverse student population,” Abramson said. 

“Dr. Pedro and I have a goal to build community through a human-first approach, and I could not be more grateful for his support and guidance in my own journey. The CRA highlights the best of what the computing discipline has to offer, and I am incredibly honored for our work to be recognized.”

Abramson will spend the summer researching how user-centered design curricula can help promote confidence, belonging, and retention for women in computing.

Nominees for the PURA program were recognized for contributing to multiple research projects, authoring or coauthoring papers, presenting at conferences, developing widely used software artifacts, and supporting their communities as teaching assistants, tutors, and mentors. 

School of Computing Instruction Communications Officer Emily Smith contributed to this story.

Main Photo: Ryan Punamiya works with a robot during the 2025 International Conference on Robotics and Automation in Atlanta. Photo by Terence Rushin/College of Computing.

Georgia Tech researchers say HPC and artificial intelligence (AI) advances this year are poised to improve how people power their homes, design safer buildings, and travel through cities.

According to Qi Tang, scientists will take progressive steps toward cleaner, sustainable energy through nuclear fusion in 2026. 

“I am very hopeful about the role of advanced computing and AI in making fusion a clean energy source,” said Tang, an assistant professor in the School of Computational Science and Engineering (CSE). 

“Fusion systems involve many interconnected processes happening across different scales. Modern simulations, combined with data-driven methods, allow us to bring these pieces together into a unified picture.”

Tang’s research connects HPC and machine learning with fusion energy and plasma physics. This year, Tang is continuing work on large-scale nuclear fusion models.

Only a few experimental fusion reactors exist worldwide compared to more than 400 nuclear fission reactors. Tang’s work supports a broader effort to turn fusion from a promising idea into a practical energy source.

Nuclear fusion occurs in plasma, the fourth state of matter, where gas is heated to millions of degrees. In this extreme state, electrons are stripped from atoms, creating a hot soup of fast-moving ions and free electrons. In plasma, hydrogen atoms overcome their natural electrical repulsion, collide, and fuse together. This releases energy that can power cities and homes.

Computers interpret extreme temperatures, densities, pressures, and plasma particle motion as massive datasets. Tang works to assimilate these data types from computer models and real-world experiments.

To do this, he and other researchers rely on machine learning approaches to analyze data across models and experiments more quickly and to produce more accurate predictions. Over time, this will allow scientists to test and improve fusion reactor designs toward commercial use. 

Beyond energy and nuclear engineering, Umar Khayaz sees broader impacts for HPC in 2026.

“HPC is the need of the day in every field of engineering sciences, physics, biology, and economics,” said Khayaz, a CSE Ph.D. student in the School of Civil and Environmental Engineering. 

“HPC is important enough to say that we need to employ resources to also solve social problems.”

Khayaz studies dynamic fracture and phase-field modeling. These areas explore how materials break under sudden, rapid loads. 

Like nuclear fusion, Khayaz says dynamic fracture problems are complex and data-intensive. In 2026, he expects to see more computing resources and computational capabilities devoted to understanding these problems and other emerging civil engineering challenges.

CSE Ph.D. student Yiqiao (Ahren) Jin sees a similar relationship between infrastructure and self-driving vehicles. He believes AI will innovate this area in 2026.

At Georgia Tech, Jin develops efficient multimodal AI systems. An autonomous vehicle is a multimodal system that uses camera video, laser sensors, language instructions, and other inputs to navigate city streets under changing scenarios like traffic and weather patterns.

Jin says multimodal research will move beyond performance benchmarks this year. This shift will lead to computer systems that can reason despite uncertainty and explain their decisions. In result, engineers will redefine how they evaluate and deploy autonomous systems in safety-critical settings.

“Many foundational problems in perception, multimodal reasoning, and agent coordination are being actively addressed in 2026. These advances enable a transition from isolated autonomous systems to safer, coordinated autonomous vehicle fleets,” Jin said. 

“As these systems scale, they have the potential to fundamentally improve transportation safety and efficiency.”

To support emergent startups, the early-stage accelerator will establish a collaborative facility at The Biltmore in Atlanta’s Tech Square, the national innovation district and dedicated area in the city that fosters community growth and meaningful innovation at the heart of the city’s tech scene. 

“Atlanta is where innovation becomes opportunity, and Velocity Startups will make that journey even faster,” said Donnie Beamer, senior technology advisor in the Atlanta Mayor’s Office of Technology and Innovation. “By connecting entrepreneurs to the critical resources they need to scale, we are fueling more startups, creating more jobs, and driving economic growth. Ultimately, this will secure Atlanta’s place as a top global destination for innovation, investment, and entrepreneurial success.”

As an early-stage accelerator, Velocity Startups provides resources — including mentorship support, space, tools, networks, and infrastructure — to Georgia Tech students, faculty, researchers, and the greater Atlanta community, bridging the gap from spinoff to viable startup. At Georgia Tech, many startups that complete the CREATE-X Startup Launch program and present at the Demo Day event will gain access to Velocity Startups. The accelerator will also offer strategic programming, funding, and access to Georgia Tech’s research resources and serve as a coordinating entity for Metro Atlanta’s entrepreneurial ecosystem, engaging more than 50 colleges and advocating for policies that support startup success. 

“Velocity Startups represents a pivotal step in bringing together the resources, expertise, and entrepreneurial spirit within our ecosystems as we look to further establish Atlanta as a top national tech hub. By uniting these elements, Velocity Startups will help startups scale from their first customer to long-term growth,” said Raghupathy “Siva” Sivakumar, vice president of commercialization and chief commercialization officer at Georgia Tech and president of Georgia Advanced Technology Ventures. “This accelerator enables the communities at Georgia Tech and beyond to translate groundbreaking research into high-impact ventures.”

Velocity Startups is a subsidiary of Georgia Advanced Technology Ventures and will operate in partnership with the City of Atlanta. A national search is currently underway for a director to lead the accelerator. 

For additional information about Velocity Startups, visit commercialization.gatech.edu/velocity.

Students and faculty from the School of Computational Science and Engineering (CSE) are leading the Georgia Tech contingent at the SIAM Conference on Computational Science and Engineering (CSE25). The Society of Industrial and Applied Mathematics (SIAM) organizes CSE25, occurring March 3-7 in Fort Worth, Texas.

At CSE25, the School of CSE researchers are presenting papers that apply computing approaches to varying fields, including:                   

• Experiment designs to accelerate the discovery of material properties
• Machine learning approaches to model and predict weather forecasting and coastal flooding
• Virtual models that replicate subsurface geological formations used to store captured carbon dioxide
• Optimizing systems for imaging and optical chemistry
• Plasma physics during nuclear fusion reactions

[Related: GT CSE at SIAM CSE25 Interactive Graphic] 

“In CSE, researchers from different disciplines work together to develop new computational methods that we could not have developed alone,” said School of CSE Professor Edmond Chow. 

“These methods enable new science and engineering to be performed using computation.” 

CSE is a discipline dedicated to advancing computational techniques to study and analyze scientific and engineering systems. CSE complements theory and experimentation as modes of scientific discovery. 

Held every other year, CSE25 is the primary conference for the SIAM Activity Group on Computational Science and Engineering (SIAG CSE). School of CSE faculty serve in key roles in leading the group and preparing for the conference.

In December, SIAG CSE members elected Chow to a two-year term as the group’s vice chair. This election comes after Chow completed a term as the SIAG CSE program director. 

School of CSE Associate Professor Elizabeth Cherry has co-chaired the CSE25 organizing committee since the last conference in 2023. Later that year, SIAM members reelected Cherry to a second, three-year term as a council member at large. 

At Georgia Tech, Chow serves as the associate chair of the School of CSE. Cherry, who recently became the associate dean for graduate education of the College of Computing, continues as the director of CSE programs. 

“With our strong emphasis on developing and applying computational tools and techniques to solve real-world problems, researchers in the School of CSE are well positioned to serve as leaders in computational science and engineering both within Georgia Tech and in the broader professional community,” Cherry said. 

Georgia Tech’s School of CSE was first organized as a division in 2005, becoming one of the world’s first academic departments devoted to the discipline. The division reorganized as a school in 2010 after establishing the flagship CSE Ph.D. and M.S. programs, hiring nine faculty members, and attaining substantial research funding.

Ten School of CSE faculty members are presenting research at CSE25, representing one-third of the School’s faculty body. Of the 23 accepted papers written by Georgia Tech researchers, 15 originate from School of CSE authors.

The list of School of CSE researchers, paper titles, and abstracts includes:
Bayesian Optimal Design Accelerates Discovery of Material Properties from Bubble Dynamics
Postdoctoral Fellow Tianyi Chu, Joseph Beckett, Bachir Abeid, and Jonathan Estrada (University of Michigan), Assistant Professor Spencer Bryngelson
[Abstract]

Latent-EnSF: A Latent Ensemble Score Filter for High-Dimensional Data Assimilation with Sparse Observation Data
Ph.D. student Phillip Si, Assistant Professor Peng Chen
[Abstract]

A Goal-Oriented Quadratic Latent Dynamic Network Surrogate Model for Parameterized Systems
Yuhang Li, Stefan Henneking, Omar Ghattas (University of Texas at Austin), Assistant Professor Peng Chen
[Abstract]

Posterior Covariance Structures in Gaussian Processes
Yuanzhe Xi (Emory University), Difeng Cai (Southern Methodist University), Professor Edmond Chow
[Abstract]

Robust Digital Twin for Geological Carbon Storage
Professor Felix Herrmann, Ph.D. student Abhinav Gahlot, alumnus Rafael Orozco (Ph.D. CSE-CSE 2024), alumnus Ziyi (Francis) Yin (Ph.D. CSE-CSE 2024), and Ph.D. candidate Grant Bruer
[Abstract]

Industry-Scale Uncertainty-Aware Full Waveform Inference with Generative Models
Rafael Orozco, Ph.D. student Tuna Erdinc, alumnus Mathias Louboutin (Ph.D. CS-CSE 2020), and Professor Felix Herrmann
[Abstract]

Optimizing Coupled Systems: Insights from Co-Design Imaging and Optical Chemistry
Assistant Professor Raphaël Pestourie, Wenchao Ma and Steven Johnson (MIT), Lu Lu (Yale University), Zin Lin (Virginia Tech)
[Abstract]

Multifidelity Linear Regression for Scientific Machine Learning from Scarce Data
Assistant Professor Elizabeth Qian, Ph.D. student Dayoung Kang, Vignesh Sella, Anirban Chaudhuri and Anirban Chaudhuri (University of Texas at Austin)
[Abstract]

LyapInf: Data-Driven Estimation of Stability Guarantees for Nonlinear Dynamical Systems
Ph.D. candidate Tomoki Koike and Assistant Professor Elizabeth Qian
[Abstract]

The Information Geometric Regularization of the Euler Equation
Alumnus Ruijia Cao (B.S. CS 2024), Assistant Professor Florian Schäfer
[Abstract]

Maximum Likelihood Discretization of the Transport Equation
Ph.D. student Brook Eyob, Assistant Professor Florian Schäfer
[Abstract]

Intelligent Attractors for Singularly Perturbed Dynamical Systems
Daniel A. Serino (Los Alamos National Laboratory), Allen Alvarez Loya (University of Colorado Boulder), Joshua W. Burby, Ioannis G. Kevrekidis (Johns Hopkins University), Assistant Professor Qi Tang (Session Co-Organizer)
[Abstract]

Accurate Discretizations and Efficient AMG Solvers for Extremely Anisotropic Diffusion Via Hyperbolic Operators
Golo Wimmer, Ben Southworth, Xianzhu Tang (LANL), Assistant Professor Qi Tang 
[Abstract]

Randomized Linear Algebra for Problems in Graph Analytics
Professor Rich Vuduc
[Abstract]

Improving Spgemm Performance Through Reordering and Cluster-Wise Computation
Assistant Professor Helen Xu
[Abstract]

Inspired by SpaceX’s Super Heavy booster, a team led by Georgia Tech’s Spencer Bryngelson and New York University’s Florian Schäfer modeled the turbulent interactions of a 33-engine rocket. Their experiment set new records, running the largest ever fluid dynamics simulation by a factor of 20 and the fastest by over a factor of four.

The team ran its custom software on the world’s two fastest supercomputers, as well as the eighth fastest, to construct such a massive model.

Applications from the simulation reach beyond rocket science. The same computing methods can model fluid mechanics in aerospace, medicine, energy, and other fields. At the same time, the work advances understanding of the current limits and future potential of computing. 

The team finished as runners-up for the 2025 Gordon Bell Prize for its impactful, multi-domain research. Referred to as the Nobel Prize of supercomputing, the award was presented at the world’s top conference for high-performance computing (HPC) research.

“Fluid dynamics problems of this style, with shocks, turbulence, different interacting fluids, and so on, are a scientific mainstay that marshals our largest supercomputers,” said Bryngelson, an assistant professor with the School of Computational Science and Engineering (CSE).

“Larger and faster simulations that enable solutions to long-standing scientific problems, like the rocket propulsion problem, are always needed. With our work, perhaps we took a big dent out of that issue.”

The Super Heavy booster reflects the space industry’s move toward reusable multi-engine first-stage rockets that are easier to transport and more economical overall. 

However, this shift creates research and testing challenges for new designs.

Each of Super Heavy’s 33 thrusters expels propellant at ten times the speed of sound. As individual engines reach extreme temperatures, pressures, and densities, their combined interactions with the airframe make such violent physics even more unpredictable.

Frequent physical experiments would be expensive and risky, so scientists rely on computer models to supplement the engineering process. 

Bryngelson’s flagship Multicomponent Flow Code (MFC) software anchored the experiment. MFC is an open-source computer program that simulates fluid dynamic models. Bryngelson’s lab has been modifying MFC since 2022 to run on more powerful computers and solve larger problems. 

In computing terms, this MFC-enhanced model simulated fluid flow resolution at 200 trillion grid points and one quadrillion degrees of freedom. These metrics exceeded previous record-setting benchmarks that tallied 10 trillion and 30 trillion grid points.

This means MFC simulations provide greater detail and capture smaller-scale features than previous approaches. The rocket simulation also ran four times faster and achieved 5.7 times the energy efficiency of comparable methods.   

Integrating information geometric regularization (IGR) into MFC played a key role in attaining these results. This new approach improved the simulation’s computational efficiency and overcame the challenge of shock dynamics.

In fluid mechanics, shock waves occur when objects move faster than the speed of sound. Along with hampering the performance of airframes and propulsion systems, shocks have historically been difficult to simulate.

Computational scientists have used empirical models based on artificial viscosity to account for shocks. Although these approaches mimic the physical effects of shock waves at the microscopic scale, they struggle to effectively capture the large-scale features of the flow. 

Information geometry uses curved spaces to study concepts of statistics and information. IGR uses these tools to modify the underlying geometry in fluid dynamics equations. When traveling in the modified geometry, fluid in the model preserves the shocks in a more natural way. 

“When regularizing shocks to much larger scales relevant in these numerical simulations, conventional methods smear out important fine-scale details,” said Schäfer, an assistant professor at NYU’s Courant Institute of Mathematical Sciences.

“IGR introduces ideas from abstract math to CFD that allow creating modified paths that approach the singularity without ever reaching it. In the resulting fluid flow, shocks never become too spiky in simulations, but the fine-scale details do not smear out either.” 

Simulating a model this large required the Georgia Tech researchers to run MFC on El Capitan and Frontier, the world's two fastest supercomputers. 

The systems are two of four exascale machines in existence. This means they can solve at least one quintillion (“1” followed by 18 zeros) calculations per second. If a person completed a simple math calculation every second, it would take that person about 30 billion years to reach one quintillion operations.

Frontier is housed at Oak Ridge National Laboratory and debuted as the world’s first exascale supercomputer in 2022. El Capitan surpassed Frontier when Lawrence Livermore National Laboratory launched it in 2024.

To prepare MFC for performance on these machines, Bryngelson’s lab followed a methodical approach spanning years of hardware acquisition and software engineering. 

In 2022, Bryngelson attained an AMD MI210 GPU accelerator. Optimizing MFC on the component played a critical step toward preparing the software for exascale machines.

AMD hardware underpins both El Capitan and Frontier. The MI300A GPU powers El Capitan while Frontier uses the MI250X GPU. 

After configuring MFC on the MI210 GPU, Bryngelson’s lab ran the software on Frontier for the first time during a 2023 hackathon. This confirmed the code was ready for full-scale deployment on exascale supercomputers based on AMD hardware. 

In addition to El Capitan and Frontier, the simulation ran on Alps, the world’s eight-fastest supercomputer based at the Swiss National Supercomputing Centre. It is the largest available system that features the NVIDIA GH200 Grace Hopper Superchip.

Like with AMD GPUs, Bryngelson acquired four GH200s in 2024 and began configuring MFC to the latest hardware innovation powering New Age supercomputers. Later that year, the Jülich Research Centre accepted Bryngelson’s group into an early access program to test JUPITER, a developing supercomputer based on the NVIDIA superchip.

The group earned a certificate for scaling efficiency and node performance on the way toward validating that their code worked on the GH200. The early access project proved successful for JUPITER, which launched in 2025 as Europe’s fastest supercomputer and fourth fastest in the world.

“Getting the level of hands-on experience with world-leading supercomputers and computing resources at Georgia Tech through this project has been a fantastic opportunity for a grad student,” said CSE Ph.D. student Ben Wilfong.

“To leverage these machines, I learned more advanced programming techniques that I’m glad to have in my tool belt for future projects. I also enjoyed the opportunity to work closely with and learn from industry experts from NVIDIA, AMD, and HPE/Cray.”

El Capitan, Frontier, JUPITER, and Alps maintained their rankings at the 2025 International Conference for High Performance Computing Networking, Storage and Analysis (SC25). Of note, the TOP500 announced at SC25 that JUPITER surpassed the exaflop threshold. 

The SC Conference Series is one of two venues where the TOP500 announces updated supercomputer rankings every June and November. The TOP500 ranks and details the 500 most powerful supercomputers in the world. 

The SC Conference Series serves as the venue where the Association for Computing Machinery (ACM) presents the Gordon Bell Prize. The annual award recognizes achievement in HPC research and application. The Tech-led team was among eight finalists for this year’s award.

Along with Bryngelson, Georgia Tech members included Ph.D. students Anand Radhakrishnan and Wilfong, postdoctoral researcher Daniel Vickers, alumnus Henry Le Berre (CS 2025), and undergraduate student Tanush Prathi.

Schäfer’s partnership with the group stems from his previous role as an assistant professor at Georgia Tech from 2021 to 2025. 

Collaborators on the project included Nikolaos Tselepidis and Benedikt Dorschner from NVIDIA, Reuben Budiardja from ORNL, Brian Cornille from AMD, and Stephen Abbot from HPE. All were co-authors of the paper and named finalists for the Gordon Bell Prize. 

“I’m elated that we have been nominated for such a prestigious award. It wouldn't have been possible without the combined and diligent efforts of our team,” Radhakrishnan said. 

“I’m looking forward to presenting our work at SC25 and connecting with other researchers and fellow finalists while showcasing seminal work in the field of computing.”

The surface and atmosphere of Mars have seen many changes over its 4.5-billion-year history. While the planet's current atmosphere is very thin (about 0.6% of Earth's), it was once thick enough to sustain liquid water.

According to new research published in Communications Earth & Environment, these atmospheric changes could play a key role in how we interpret sediment deposits on the planet.

“We found that the changing pressure resulting from atmospheric changes would have produced sediment-rich water flows with varying shapes over time,” says co-author and Georgia Tech Assistant Professor Frances Rivera-Hernández, adding that since Mars’ present-day atmosphere is very thin, the associated low pressures would produce behaviors not seen on Earth. 

“Earth’s thicker atmosphere means that there are higher pressures on our planet, which produce very different behaviors,” she explains. “This means that Earth analogs may not be reliable for interpreting some Martian sedimentary landscapes.”

“At low present-day pressures, Mars mud would boil and levitate if the surface temperature was warm, or freeze and flow more like lava if the temperature was cold,” adds study lead Jacob Adler, who began working on the project while a postdoctoral researcher in Rivera-Hernández’s PLANETAS Lab at Georgia Tech, and continued the study in his current role as an assistant research professor in Arizona State University's School of Earth and Space Exploration. 

The team also included Georgia Tech Ph.D. student and current PLANETAS Lab member Sharissa Thompson, along with researchers from the Open University and Czech Academy of Sciences.

“This study adds a critical layer of nuance to analogue research,” says Rivera-Hernández. “By comparing our lab results to real Martian landforms, we can better reconstruct Mars’ past climate — leading to increasingly successful research in the future.”

Making Martian mud

In order to recreate past conditions on the red planet, the team conducted over 70 experiments in a Mars simulation chamber, testing how flowing water-sediment mixtures would be affected by the varying pressures and temperatures throughout the planet’s history.

Thompson, who specializes in understanding these types of mixtures, played a key role in interpreting the results. “As part of my Ph.D. work at Georgia Tech, I uncover how and why flow shapes evolve as pressure changes, which helped us understand how these flows could have shifted with changing pressures on Mars over time,” she says. “I’m thrilled to have contributed to the innovative flow experiments this study conducted.”

The experiments revealed that at higher atmospheric pressures, water and mud would have similar flow physics (rheology) as on Earth, indicating that some of the oldest sedimentary features on the surface should appear similar to Earth environments. In these scenarios, surface conditions may also have been more habitable for life.

On the other hand, as Mars started to lose most of its atmosphere, the dominant physics in sediment flow experiments changed to freezing and boiling. The team found that at the lower pressures Mars has experienced after the Noachian, the rheology and deposit shapes (morphology) were not at all Earth-like.

“When we mapped out where on Mars, we would expect this different behavior, we found that this opposite behavior could happen at the same time at different locations on the planet,” Adler shares. “The small-scale climate variations across Mars’ topography are enough to see these opposing effects.”

Decoding Mars' past

The research suggests that studying the specific shapes of features like sediment flows, debris flows and mudflows could help scientists better estimate climate conditions. It also highlights how laboratory experiments are a critical part of planetary science activities, as they can help scientists better interpret remote sensing and modeling results.

"By finding matching morphologies of what we see on Mars and what we see in these lab experiments, we might be able to better time-stamp the paleoclimate record,” Adler explains.

"We’ve sent rover missions to Mars largely because we find compelling remote sensing evidence of deposits formed by water or mud that could indicate a habitable environment,” he adds. “We are often eager to compare what we find to Earth analogs, but these are not always suitable for comparison. This study shows there is still much we can learn about Mars by conducting experiments under Mars conditions.”

Funding: NASA

DOI: https://doi.org/10.1038/s43247-025-02879-w 

311 systems allow residents to report potholes, broken fire hydrants, and other municipal issues. In recent years, the use of artificial intelligence (AI) to provide 311 services to community residents has boomed across city and state governments. This includes an artificial virtual assistant (AVA) developed by third-party vendors for the City of Atlanta in 2023.

Through survey data, researchers from Tech’s School of Interactive Computing found that many residents are generally positive about 311 chatbots. In addition to eliminating long wait times over the phone, they also offer residents quick answers to permit applications, waste collection, and other frequently asked questions.

However, the study, which was conducted in Atlanta, indicates that 311 chatbots could be causing residents to feel isolated from public officials and less aware of what’s happening in their community.

Jieyu Zhou, a Ph.D. student in the School of IC, said it doesn’t have to be that way.

Uniting Communities

Zhou and her advisor, Assistant Professor Christopher MacLellan, published a paper at the 2025 ACM Designing Interactive Systems (DIS) Conference that focuses on improving public service chatbot design and amplifying their civic impact. They collaborated with Professor Carl DiSalvo, Associate Professor Lynn Dombrowski, and graduate students Rui Shen and Yue You.

Zhou said 311 chatbots have the potential to be agents that drive community organization and improve quality of life.

“Current chatbots risk isolating users in their own experience,” Zhou said. “In the 311 system, people tend to report their own individual issues but lose a sense of what is happening in their broader community. 

“People are very positive about these tools, but I think there’s an opportunity as we envision what civic chatbots could be. It’s important for us to emphasize that social element — engaging people within the community and connecting them with government representatives, community organizers, and other community members.”

Zhou and MacLellan said 311 chatbots can leave users wondering if others in their communities share their concerns.

“If people are at a town hall meeting, they can get a sense of whether the problems they are experiencing are shared by others,” Zhou said. “We can’t do that with a chatbot. It’s like an isolated room, and we’re trying to open the doors and the windows.”

Adding a Human Touch

In their paper, the researchers note that one of the biggest criticisms of 311 chatbots is they can’t replace interpersonal interaction.

Unlike chatbots, people working in local government offices are likely to:

• Have direct knowledge of issues
• Provide appropriate referrals
• Empathize with the resident’s concerns

MacLellan said residents are likely to grow frustrated with a chatbot when reporting issues that require this level of contextual knowledge.

One person in the researchers’ survey noted that the chatbot they used didn’t understand that their report was about a sidewalk issue, not a street issue.

“Explaining such a situation to a human representative is straightforward,” MacLellan said. “However, when the issue being raised does not fall within any of the categories the chatbot is built to address, it often misinterprets the query and offers information that isn’t helpful.”

The researchers offer some design suggestions that can help chatbots foster community engagement and improve community well-being:

• Escalation. Regarding the sidewalk report, the chatbot did not offer a way to escalate the query to a human who could resolve it. Zhou said that this is a feature that chatbots should have but often lack.
• Transparency. Chatbots could provide details about recent and frequently reported community issues. They should inform users early in the call process about known problems to help avoid an overload of user complaints.
• Education. Chatbots can keep users updated about what’s happening in their communities.
• Collective action. Chatbots can help communities organize and gather ideas to address challenges and solve problems.

“Government agencies may focus mainly on fixing individual issues,” Zhou said, “But recognizing community-level patterns can inspire collective creativity. For example, one participant suggested that if many people report a broken swing at a playground, it could spark an initiative to design a new playground together—going far beyond just fixing it.”

These are just a few examples of things, the researchers argue, that 311 services were originally designed to achieve.

“Communities were already collaborating on identifying and reporting issues,” Zhou said. “These chatbots should reflect the original intentions and collaboration practices of the communities they serve.

“Our research suggests we can increase the positive impact of civic chatbots by including social aspects within the design of the system, connecting people, and building a community view.”

School of Electrical and Computer Engineering (ECE) Ph.D. students Gus Richter, Malhar Tamhane, and Felipe Sandoval are the latest to make the trip to the “Last Frontier” as they work to push the boundaries of atmospheric research. The trio participated in the 2025 Polar Aeronomy and Radio Science (PARS) summer school program held in August at the University of Alaska Fairbanks and the High-frequency Active Auroral Research Program (HAARP).

Read the full story on the School of Electrical and Computer Engineering's website.

The course, BIOS 4590, is a research project lab for senior biology majors that provides an opportunity for professors to share their expertise with students in a hands-on environment. In his class, Associate Professor Vinayak (Vinny) Agarwal, who holds joint appointments in the School of Chemistry and Biochemistry and School of Biological Sciences, aimed to introduce undergraduates to advanced bioinformatics tools through applied research using new-to-science raw data. 

The resulting paper, “Phylogenomic Identification of a Highly Conserved Copper-Binding RiPP Biosynthetic Gene Cluster in Marine Microbulbifer Bacteria,” which was recently published in ACS Chemical Biology, involves the historically understudied genus of Microbulbifer, a type of bacteria often associated with sponges and corals. These microbial communities are rich sources of natural products, small biological molecules often associated with medicine and drug discovery. 

"This class, and the resulting research, is a testament to the transformative power of hands-on learning,” says Susan Lozier, dean of the College of Sciences, Betsy Middleton and John Clark Sutherland Chair, and professor in the School of Earth and Atmospheric Sciences. “The success of this course — and the students’ remarkable achievement — reflects Georgia Tech's commitment to fostering curiosity, collaboration, and scientific rigor and to empowering the next generation of scientists and leaders."

Funded by Agarwal’s 2023 National Science Foundation CAREER grant and Camille and Henry Dreyfus Foundation Teacher-Scholar award, the class also received support from leadership in the College of Sciences, School of Biological Sciences, and School Chemistry and Biochemistry. The study’s lead author, graduate student Yifan (Grace) Tang, served as the class teaching assistant, and was funded in part by a Biochemistry and Biophysics Graduate Assistance in Areas of National Need fellowship. 

“The students in this class are working on important, novel work — this cohort worked with real genomic data that had never been sequenced before,” she says. “Typically, researchers might work with one or two genome sequences, but we provided students with 42 — this might be the first time anyone has looked at Microbulbifer at such a wide scope.” 

From classroom to publication

To prepare for the class, Tang worked alongside Laboratory Manager Alison Onstine, who manages the School of Biological Sciences teaching laboratory spaces, to sequence the Key West bacterial genomes.

“Our work in the Agarwal Lab is in natural product discovery. We focus on finding new pharmaceutical drugs through marine bacteria — but with a bioinformatics spin,” Tang explains. “We wanted to bring this type of experience to undergraduates, so we gave fully sequenced genomes to students and asked them to look for potential properties.” 

Throughout the class, students learned different techniques for analyzing bacterial genome sequences and extracting data with various tools — gaining both lab and computational skills through hands-on experiences, live demos, and troubleshooting sessions. 

“The highlight was showing students just how much we can learn about a bacterial genus, especially one that hasn’t been studied at this scale before,” Tang shares. “This is a growing field, so there are so many opportunities for students to make meaningful contributions while learning new skills.”

Empowering future students

For many students, it was their first time using these types of tools, but Agarwal says that it’s something they'll likely encounter in both industry and research. He sees this type of research experience as especially helpful for seniors, who are often deciding between entering the workforce or continuing their education.

“Bioinformatics is increasingly important for analyzing big data. Students need the ability to manipulate and understand data using computational tools, and this class plays an important role in familiarizing them with this process,” he shares. “Our goal is to demystify research and give students the confidence and tools for both graduate school and for the workforce after graduation.”

The class will be offered for a third time in Fall 2026. While the exact course of research hasn’t yet been decided, “we always aim for something new that can produce publication-quality research — students don’t repeat past year’s work,” Agarwal says. This recent cohort of students built on the success of 18 undergraduates who took the class in 2023, who also published a paper. “This course truly underscores Georgia Tech’s commitment to pioneering meaningful undergraduate experiences — no other peer institution I know of is exposing undergraduates to bioinformatics at this level.”

Funding: NSF CAREER and the Dreyfus Foundation

Matsumoto, who is also a principal investigator at the International Institute for Sustainability with Knotted Chiral Meta Matter (WPI-SKCM2) at Hiroshima University, is the corresponding author on a new study exploring the physics of ‘jamming’ — a phenomenon when soft or stretchy materials become rigid under low stress but soften under higher tension.

The study, "Pulling Apart the Mechanisms That Lead to Jammed Knitted Fabrics," was published this week in Physical Review E, and also includes Georgia Tech Matsumoto Group graduate students Sarah Gonzalez and Alexander Cachine in addition to former postdoctoral fellow Michael Dimitriyev, who is now an assistant professor at Texas A&M University.

The work builds on the group’s previous research demonstrating that knitted materials can be mathematically ‘programmed’ to behave in predictable ways. “These properties are intuitively understood by people who knit by hand,” Matsumoto says, “but in order to manipulate and use these behaviors in an industrial setting, we need to understand the physics behind them. This new research is another step in that direction.”

An Unexpected Twist

Gonzalez, who led the research, first became interested in jamming while conducting adjacent research. “I was using model simulations to characterize how different yarn properties affect the behavior of knitted fabrics and noticed a strange stiff region,” she recalls. “In our previous research, we had also seen this behavior in lab experiments, which suggested that what we were seeing in the simulations was a genuine phenomenon. I wanted to investigate it further.”

After digging into the topic, she realized that what she was seeing was called ‘jamming.’ In knits, Gonzalez explains, jamming occurs when stitches are packed tightly together, and the fabric resists stretching. Although it’s a well-known phenomenon, the physics has mostly been investigated in granular systems, like snow or sand, rather than fabrics.

“In fabrics, when you pull softly, the response is surprisingly stiff, but when you start pulling harder and harder, the stitches rearrange, and the material softens,” Matsumoto says. “In granular systems, this is a little like how avalanches work. At low forces, the snow pack is solid, but when the slope is steep, the force of gravity liquidizes that snow pack into an avalanche.”

“In fabrics, it is a little like having a tangle in a piece of jewelry,” she adds. “If you pull on it, it gets quite stiff, but if you loosen the knot, the chain can reconfigure, and it's not so stiff.”

Unraveling the Physics of Jamming

Using a combination of experiments with industrially knitted fabrics and computer models, the team analyzed what causes jamming in fabrics and how to control it. “We wanted to determine how different yarn properties impacted jamming,” Gonzalez explains. “Our goal was to understand the mechanics of jamming through how yarn interacts at various touchpoints in stitches.”

The team found that both machine tension and yarn thickness played a key role in making a fabric more or less jammed, and that jamming behaves differently depending on which direction the fabric is stretched.

“When you stretch a knit along the rows, the stiffness of the yarn causes fabric jamming. Jamming in the other direction is due to yarn contacts,” says Gonzalez. “We also showed that the impacts of changing machine tension and yarn thickness differ depending on fabric direction.” 

“Discovering that fabric jamming works differently in different directions was a key insight,” she adds. “To our knowledge, the physics of this has never been explored before.”

Modern Innovation — With a Centuries-Old Technique

The research dovetails with Matsumoto’s WPI-SKCM2 Center work, which involves investigating fundamental aspects of knots and chirality. The Center is interested in a class of materials called “knotted chiral meta matter” that could lead to more sustainable materials.

For example, knitting — which leverages chiral knots — could be used to create more elastic fabrics from natural materials. “In many cases, manufacturers use yarns that combine, for example, polyester, cotton, and elastane to create a desired elasticity,” Matsumoto says. “Our research suggests that manipulating the topology of the stitches could lead to a similar elasticity, reducing the need for petroleum-based fibers and creating a more sustainable textile.”

“Knitting has the potential to be extremely useful in manufacturing, but knowledge has typically been shared through intuition and word of mouth,” she adds. “By creating these mathematical models, we hope to formalize that knowledge in a way that’s accessible for large-scale manufacturing — so we can leverage this centuries-old intuition for modern innovation.”

Funding: This work was supported by the World Premier International Research Center Initiative (WPI), Ministry of Education, Culture, Sports, Science and Technology of Japan; National Science Foundation (NSF); and Research Corporation for Science Advancement (RCSA).

DOI: https://doi.org/10.1103/g94g-c6tt 

Georgia Tech roboticists are attempting to identify the most stressful periods that human teleoperators experience while performing tasks remotely. A novel study provides new insights into determining when a teleoperator needs to operate at a high level of focus and which parts of the task can be delegated to robot automation.

School of Interactive Computing Associate Professor Matthew Gombolay calls it the “sweet spot” of human ingenuity and robotic precision. Gombolay and students from his CORE Robotics Labconducted a novel study that measures stress and workload on human teleoperators.

Gombolay said it can inform military officials on how to strategically implement task automation and maximize human teleoperator performance.

Humans continue to hand over more tasks to robots to perform, but Gombolay said that some functions will still require human input and oversight for the foreseeable future.

Specific applications, such as space exploration, commercial and military aviation, disaster relief, and search and rescue, pose substantial safety concerns. Astronauts stationed on the International Space Station, for example, manually control robots that bring in supplies, move cargo, and make structural repairs.

“It’s brutal from a psychological perspective,” Gombolay said.

The question often asked about automating a task in these fields is, at what point can a robot be trusted more than a human?

A recent paper by Gombolay and his current and former students — Sam Yi Ting, Erin Hedlund-Botti, and Manisha Natarajan — sheds new light on the debate. The paper was published in the IEEE Robotics and Automation Letters and will be presented at the International Conference on Robotics and Automation in Atlanta.

The NASA-funded study can identify which aspects of tedious, time-consuming tasks can be automated and which require human supervision. If roboticists can pinpoint the elements of a task that cause the least stress, they can automate these components and enable humans to oversee the more challenging aspects.

“If we’re talking about repetitive tasks, robots do better with that, so if you can automate it, you should,” said Ting, a former grad student and lead author of the paper. “I don’t think humans enjoy doing repetitive tasks. We can move toward a better future with automation.”

Military officials, for example, could measure the stress of remote drone pilots and know which times during a pilot’s shift require the highest level of attention.

“We can get a sense of how stressed you are and create models of how divided your attention is and the performance rate of the tasks you’re doing,” Gombolay said.

“It can be a low-stress or high-stress situation depending on the stakes and what’s going on with you personally. Are you well-caffeinated? Well-rested? Is there stress from home you’re bringing with you to the workplace? The goal is to predict how good your task performance will be. If it indicates it might be poor, we may need to outsource work to other people or create a safe space for the operator to destress.”

The Stress Test

For their study, the researchers cut a small river-shaped path into a medium-density fiberboard. The exercise required the 24 participants to use a remote robotic arm to navigate through the path from one end to the other without touching the edges.

The experiment grew more challenging as new stress conditions and workload requirements were introduced. The changing conditions required the test participants to multitask to complete the assignment.

Gombolay said the study supports the Yerkes-Dodson Law, which states that moderate levels of stress increase human performance.

The experiment showed that operators felt overwhelmed and performed poorly when multitasking was introduced. Too much stress led to poor performance, but a moderate amount of stress induced more engagement and enhanced teleoperator focus. 

Ting said finding that ideal stress zone can lead to a higher performance rating. 

“You would think the more stressed you are, the more your performance decreases,” Ting said. “Most people didn’t react that way. As stress increased, performance increased, but when you increased workload and gave them more to do, that’s when you started seeing deteriorating performance.”

Gombolay said no stress can be just as detrimental as too much stress. Performing a task without stress tends to cause teleoperators to become disinterested, especially if it is repetitive and time-consuming.

“No stress led to complacency,” Gombolay said. “They weren’t as engaged in completing the task.

“If your excitement is too low, you get so bored you can’t muster the cognitive energy to reason about robot operation problems.”

The Human Factor

Roboticists have made significant leaps in recent years to remove teleoperators from the equation. Still, Gombolay said it’s too early to tell whether robots can be trusted with any task that a human can perform.

“We’re a long way from full autonomy,” he said. “There’s a lot that robots still can’t do without a human operator. Search and rescue operations, if a building collapses, we don’t have much training data for robots to go through rubble by themselves to rescue people. There are ethical needs for humans to be able to supervise or take direct control of robots.”

Launched in 2016, the project was supported by the Institute’s Sustainable Communities Summer Internship Program, run by the Center for Serve-Learn-Sustain (now the Center for Sustainable Communities Research and Education), in which students work full time with community partners across Atlanta and Georgia.

Beginning in 2017, trusted advisers contributed to the success of this work, including Vernon E. Jordan Jr., Christopher Lawton, Ann McCleary and G. Wayne Clough. Clough, who served as Georgia Tech’s president from 1994 to 2008, long advocated for public service, community-engaged research, and interdisciplinary teaching and learning.

In 2019, Georgia Tech students and participating interns Brice Minix and Nabil Patel combed through decades of local newspapers, digitized school board records, and conducted interviews with community members who lived in Coffee County during desegregation. In 2020, Kara Vaughan Adams and Bennett Bush transcribed countless interviews. Samina Patel’s contributions in 2020 and 2021 included graphic and web design.

All their work laid the foundation for two virtual museum exhibits: emergingVOICES of Coffee County and Overcoming Segregation: A Journey Through Coffee County’s Forgotten Stories. The latter received the 2023 Award of Excellence from the American Association of State and Local History. Further recognition came this year when the project earned the 2025 Georgia Association of Museums’ Special Project Award for the PLAYBACK & FASTFORWARD seminar series.

T. Cat Ford, Project Director said, “The Serve-Learn-Sustain interns we partnered with from Georgia Tech were all graduates of Coffee High School. Their efforts turbo-charged our work—not only because they worked tirelessly but also because, as they preserved their own history, they offered valuable insights into their lived experience of this legacy.

Click here to learn more about SCoRE’s Sustainable Communities Internship Program.

The study, “TopoNets: High-Performing Vision and Language Models With Brain-Like Topography,” was awarded a spotlight at this year’s International Conference on Learning Representations (ICLR), a distinction given to only 2 percent of papers. The research was led by graduate student Mayukh Deb alongside School of Psychology Assistant Professor Apurva Ratan Murty.

Thirty-two of Tech’s computing, engineering, and science faculty represented the Institute at ICLR 2025, which is globally renowned for sharing cutting-edge research. 

“We started with this idea because we saw that AI models are unstructured, while brains are exquisitely organized,” says first-author Deb. “Our models with internal structure showed more than a 20 percent boost in efficiency with almost no performance losses. And this is out-of-the-box — it’s broadly applicable to other models with no extra fine-tuning needed.”

For Murty, the research also underscores the importance of a rapidly growing field of research at the intersection of neuroscience and AI. “There's a major explosion in understanding intelligence right now,” he says. “The neuro-AI approach is exciting because it helps emulate human intelligence in machines, making AI more interpretable.”

“In addition to advancing AI, this type of research also benefits neuroscience because it informs a fundamental question: Why is our brain organized the way it is?,” Deb adds. “Making AI more interpretable helps everyone.”

Brain-inspired blueprints

In the brain, neurons form topographic maps: neurons used for comparable tasks are closer together. The researchers applied this concept to AI by organizing how internal components (like artificial neurons) connect and process information. 

This type of organization has been tried in the past but has been challenging, Murty says. “Historically, rules constraining how the AI could structure itself often resulted in lower-performing models. We realized that for this type of biophysical constraint, you simply can’t map everything — you need an algorithmic solution.”

“Our key insight was an algorithmic trick that gives the same structure as brains without enforcing things that models don't respond well to,” he adds. “That breakthrough was what Mayukh (Deb) worked on.” 

The algorithm, called TopoLoss, uses a loss function to encourage brain-like organization in artificial neural networks, and it is compatible with many AI systems capable of understanding language and images. 

“The resulting training method, TopoNets, is very flexible and broadly applicable,” Murty says. “You can apply it to contemporary models very easily, which is a critical advancement when compared to previous methods.” 

Neuro-AI innovations

Murty and Deb plan to continue refining and designing brain-inspired AI systems. “All parts of the brain have some organization — we want to expand into other domains,” Deb says. “On the neuroscience side of things, we want to discover new kinds of organization in brains using these topographic systems.”

Deb also cites possibilities in robotics, especially in situations like space exploration where resources are limited. “Imagine running a model inside a robot with limited power,” he says. “Structured models can help us achieve 80 percent of performance with just 20 percent of energy consumption, saving valuable energy and space. This is still experimental, but it's the direction we are interested in exploring.”

“This success highlights the potential of a new approach, designing systems that benefit both neuroscience and AI — and beyond,” Murty adds. “We can learn so much from the human brain, and this project shows that brain-inspired systems can help current AI be better. We hope our work stimulates this conversation.”

Armed with a desire to do good in the world, and growing expertise in their current studies at the colleges of Business, Computing, Engineering, and Industrial and Systems Engineering, the students discovered how powerful interdisciplinary collaboration and cutting-edge technology can be in creating social change.

The Law, Data, and Design Lab is the brainchild of Alexander, who from a young age felt a call to serve her community. “I went to law school because I saw law as a tool to look beyond myself and contribute to the greater good,” said Alexander. “I see this as part of my purpose. Being at a public university, I take seriously the responsibility to ensure my research is outward facing, that it reaches beyond academia and helps make the world a better place.”

Read More

“Malaria kills over 500,000 annually with the mortality rate substantially higher in Africa,” says Frooman. “Our research explores how specific peptides bind to proteins that trigger immune responses.”

Frooman originally hoped the research would help her learn how to think like a scientist and gain basic lab knowledge.

She gained those skills and more, quickly becoming recognized as an exceptional researcher.

“Marielle is one of the most passionate and talented undergraduate researchers I have ever worked with,” says Andrew McShan, McShan Lab principal investigator and associate professor in the School of Chemistry and Biochemistry. “She is also a caring mentor and motivated future leader who wants to change the world. Her malaria research has the potential to provide real therapeutic outcomes, including better designs for vaccines and immunotherapy.” 

From curiosity to contribution

Frooman’s journey into undergraduate research began with persistence. After a year and a half of searching for lab opportunities, she attended a School of Chemistry and Biochemistry research showcase. She approached several graduate students and professors with no success, until she met McShan.

“Our first meeting was so relaxed and friendly that I didn’t even realize Professor McShan was the principal investigator,” admits Frooman. “That’s how it all started.”

Once she officially joined the lab, Frooman contributed to every stage of the research, including designing experiments, performing computational and wet lab work, analyzing data, and writing and presenting the paper.

Lessons in resilience

The team faced several challenges.

“The research was delayed by failure after failure,” says Frooman. “But each setback taught us something valuable.”

The team’s biggest challenge involved trying to grow crystals of the peptide/HLA (protein) complexes to determine how they fit together. They spent two years attempting various methods, but nothing worked.

Guided by McShan, Frooman and the team then came up with the idea of using computational modeling to enable a deeper understanding of how the peptides and proteins interact at both biophysical and structural levels.

“Utilizing the computational modeling enabled us to see the best bindings and turned into a game-changing insight for our research, potentially leading to the design of more effective malaria treatments and vaccines,” explains Frooman.

She is quick to credit Georgia Tech and McShan for providing her with such a valuable learning experience.

“At many universities, undergraduates rarely do meaningful research, but at Tech, it’s a priority,” explains Frooman. “I’m extremely grateful for the opportunity to grow in such a supportive environment, and to learn from mentors like Professor McShan who lead by example and make time for every student.”

Her advice to other undergraduates entering research?

“Embrace your failures. They make the successes even more rewarding,” shares Frooman.

Outside the lab

On campus, Frooman is president of the Student Affiliates of the American Chemical Society and Cleanup Crew at GT, a member of Alpha Phi International Fraternity, and a campus tour guide who serves on their executive board. 

She especially loves being a tour guide as it allows her to share her love of Georgia Tech and its people:

“Everyone is unapologetically themselves and fully invested in their major or interests. As someone who loves chemistry, I enjoy being surrounded by people who are just as dedicated to their passions.”

Frooman is a recipient of the Chance Family Scholarship, presented to two School of Chemistry and Biochemistry upperclassmen, recognizing their academic excellence, research contributions, and potential for career success in the field.

Recently, she shifted her research focus to organic synthetic chemistry and now works in the Gutekunst Lab. Her career goals include earning a Ph.D. in Chemistry with an emphasis on natural product synthesis, the lab-based creation of complex chemical compounds found in nature.

“I’ve seen what university labs can do,” says Frooman. “I hope to one day lead my own lab, advancing impactful research and mentoring the next generation of scientists.”

The study, “Evidence for a composite volcano on the rim of Jezero crater on Mars,” was published this May in the Nature-family journal Communications Earth & Environment, and underscores how much we have left to learn about one of the most well-studied regions of Mars.

Lead author Sara C. Cuevas-Quiñones completed the research as an undergraduate during a summer program at Georgia Tech; she is now a graduate student at Brown University. The team also included corresponding author Professor James J. Wray (School of Earth and Atmospheric Sciences), Assistant Professor Frances Rivera-Hernández (School of Earth and Atmospheric Sciences), and Jacob Adler, then a postdoctoral fellow at Georgia Tech and now an assistant research professor at Arizona State University. 

“Volcanism on Mars is intriguing for a number of reasons — from the implications it has on habitability, to better constraining the geologic history,” Wray says. “Jezero Crater is one of the best studied sites on Mars. If we are just now identifying a volcano here, imagine how many more could be on Mars. Volcanoes may be even more widespread across Mars than we thought.”

A mountain in the margins

Wray first noticed the mountain in 2007, while considering Jezero Crater as a graduate student. 

“I was looking at low-resolution photos of the area and noticed a mountain on the crater’s rim,” he recalls. “To me, it looked like a volcano, but it was difficult to get additional images.” At the time, Jezero Crater was newly discovered, and imaging focused almost entirely on its intriguing water history, which is on the opposite side of the 28-mile-wide crater.

Then, Jezero Crater, due to these lake-like sedimentary deposits, was selected as the landing spot for the 2020 Perseverance Rover — an ongoing NASA mission seeking signs of ancient Martian life and collecting rock samples for possible return to Earth.

However, after landing, some of the first rocks Perseverance encountered were not the sedimentary deposits one might expect from a previously-flooded area — they were volcanic. Wray suspected he might know the origin of these rocks, but to make a case for it, he would need to show that the mountain on the edge of Jezero Crater could indeed be a volcano.

A new researcher — and old data

The opportunity presented itself several months after Perseverance landed when Cuevas-Quiñones applied to a Summer Research Experience for Undergraduates (REU) program hosted by the School of Earth and Atmospheric Sciences to work with Wray. 

“A previous study led by Briony Horgan (professor of planetary science at Purdue University) had also suggested that Jezero Mons could be volcanic,” Cuevas-Quiñones says. “I began wondering if there was a way to home in on these suspicions.”

The team partnered with study coauthor Rivera-Hernández, who specializes in characterizing the surface of planets and their habitability. They decided to use datasets gathered from spacecraft orbiting Mars to compare the properties of Jezero Mons to other, known, volcanoes. “We can’t visit Mars and definitively prove that Jezero Mons is a volcano, but we can show that it shares the same properties with existing volcanoes — both here on Earth and Mars,” Wray explains.

“We used data from the Mars Odyssey Orbiter, Mars Reconnaissance Orbiter, ExoMars Trace Gas Orbiter, and Perseverance Rover, all in combination to puzzle this out,” he adds. “I think this shows that these older spacecraft can be extremely valuable long after their initial missions end — these old spacecraft can still make important discoveries and help us answer tricky questions.”

For Cuevas-Quiñones, it also underscores the importance of REU programs and opportunities for undergraduates. “I was an undergraduate student at the time, and this was my first time conducting research,” she says. “It was fascinating to learn how different data sets could be used to decode the origin of a landscape. After Jezero Mons, it became clear to me that I would continue to study Mars and other planetary bodies.”

The search for life — and determining Mars’ age

The discovery makes the crater even more intriguing in the search for past life on Mars. A volcano so close to watery Jezero Crater could add a critical source of heat on an otherwise cold planet, including the potential for hydrothermal activity — energy that life could use to thrive. 

This type of system also holds interest for Mars as a whole. “The coalescence of these two types of systems makes Jezero more interesting than ever,” shares Wray. “We have samples of incredible sedimentary rocks that could be from a habitable region alongside igneous rocks with important scientific value.” If returned to Earth, igneous rocks can be radioisotope dated to know their age very precisely. Dating the Jezero Crater samples could be used to calibrate age estimates, providing an unprecedented window into the geologic history of the planet.

The take home message? “Mars is the best place we have to look in our solar system for signs of life, and thanks to the Perseverance Rover collecting samples in Jezero, the United States has samples from the best rocks in the best place on Mars,” Wray says. “If these samples are returned to Earth, we can do incredible, groundbreaking science with them.”

DOI: https://doi.org/10.1038/s43247-025-02329-7

Funding: Cuevas-Quiñones was supported by Georgia Tech’s 2021 Research Experience for Undergraduates program sponsored by NSF and 3M corporation. Wray was supported by NASA funding for Co-Investigators on HiRISE and CaSSIS. CaSSIS is a project of the University of Bern and funded through the Swiss Space Office via ESA’s PRODEX program. The instrument hardware development was also supported by the Italian Space Agency (ASI) (ASI-INAF agreement 2020-17-HH.0), INAF/Astronomical Observatory of Padova, and the Space Research Center (CBK) in Warsaw. Support from SGF (Budapest), the University of Arizona Lunar and Planetary Lab, and NASA are also gratefully acknowledged. Operation support from the UK Space Agency is also acknowledged.

This year’s cohort spans a wide range of industries, including artificial intelligence, defense, healthcare, gaming, sustainability, media management, agriculture tech, fashion tech, education, and more. 

“These founders are in the messy middle and that's a beautiful place to be. There’s a lot of freedom in that,” said Margaret Weniger, director of Startup Launch. “We’re all going to be in this together. It's a safe space to try new things. It’s OK if it doesn't work out because what we want founders to learn is an entrepreneurial mindset and entrepreneurial spirit — something you take with you no matter what you do after this.”

Over the next 12 weeks, teams will validate ideas, build products, and acquire customers with the help of dedicated coaches, a robust founder community, and a network of mentors and alumni. 

Raghupathy "Siva" Sivakumar, Georgia Tech’s inaugural vice president of Commercialization and the faculty founder of CREATE-X, spoke about the core of CREATE-X and what it would take for founders to succeed.

“Startup Launch is not about Georgia Tech gaining from your success. We are here just for one reason, which is to make you successful,” he said. “You need to hold yourself accountable. You need to be ambitious in terms of how big a problem you solve. You need to be emphatic that the customer matters. The successful teams are 100% behind what's going to make the lives of customers easier and better.”

In 2014, CREATE-X was co-founded by Sivakumar, Steve McLaughlin(who is now the president of The Cooper Union for the Advancement of Science and Art), and other Georgia Tech faculty, including Ray Vito, Craig Forest, and Ravi Bellamkonda (who is now the executive vice president and provost of The Ohio State University). The program received its initial major philanthropic support from Chris Klaus, a Georgia Tech alumnus and tech entrepreneur, whose gift helped launch the initiative, and , played a key role in building out the program's maker courses. Over the years, CREATE-X has continued to grow, thanks largely to the philanthropic support of alumni and foundations who believe in its mission.

In the last decade, the program has produced over 650 startups, $2.4 billion in portfolio valuation, and had eight founders named to Forbes’ 30 Under 30. Wagner shared stories of past teams who pivoted dramatically — from a glucose-monitoring pillow to a sobriety app now valued at over $350 million, and from a camping gear delivery service to a billion-dollar logistics platform. 

“We don’t know which ideas will become the next unicorns,” Weniger said. “But we’re betting on you.”

At the kickoff event, McLaughlin and Klaus were honored for their contributions to Georgia Tech’s entrepreneurial ecosystem. McLaughlin encouraged the founders through the story of CREATE-X.

“From the very beginning, we challenged CREATE-X to be a startup as well. To this day, CREATE-X has raised its own money to do this. It's a reminder of what it takes to make this happen,” he said. “This is the most difficult challenge you have ever taken. I think at the time, we were probably skeptical about whether students could do it. Now we know that you can.”

Georgia Tech President Ángel Cabrera reflected on the impact of McLaughlin, Klaus, and others who saw the vision of Georgia Tech being an entrepreneurial campus. 

“Ten years ago, this was a crazy, absurd idea,” he said. “Now, 150 teams are working on their own crazy ideas. Even though sometimes there's this idea of the entrepreneur as a loner, what you learn very quickly is entrepreneurship is a team sport.”

Klaus spoke about people collaborating and helping solve problems together. 

“I'm especially inspired by Georgia with its complex history,” he said. “It continues to be a place where peace can be envisioned and pursued. I think this recognition strengthens my commitment to building bridges, resolving conflict, and lifting up voices that seek unity. As you build your businesses, you'll be building collaborations and partnerships, and hopefully make the world a better place.”

As the summer progresses, founders will be guided by CREATE-X’s core values: experiential education, entrepreneurial confidence, and real-world impact. Weniger encouraged teams to “show up uncomfortable” and “leverage every single resource” available.

The journey will culminate at Demo Day, where teams will showcase their startups to investors, industry leaders, and the broader community. The event is free, open to the public, and promises a front-row seat to the next wave of Georgia Tech-born innovation.

Demo Day 2025 will take place on Thursday, Aug. 28, at 5 p.m., in the Exhibition Hall. For more information and to RSVP, visit the CREATE-X Demo Day Eventbrite.

She said the police officer who interviewed her was dismissive and neglected to write down details that Foriest found significant. The deficient police report was picked up by local media, which led to news stories that inaccurately described the crime and left out important information.

Foriest said she learned from the incident that incomplete information doesn’t mitigate violence. The perspectives and stories of people who experience violence are essential to reliable data.

The incident guided Foriest as she committed to research that gathers complete and accurate data on multiple types of violence, including violent injury and homicide, intimate partner violence, gender-based violence, and suicide.

Foriest earned a bachelor’s in health science from Columbus State University. She also holds two master’s degrees: one in public health from the University of Southern California, and another in technology leadership and management from Agnes Scott College.

In 2021, Foriest started her Ph.D. in human-centered computing at Georgia Tech to understand how technology influences violence.

“I look at all types of violence as an outcome of how technology affects communication,” she said. 

One thing she discovered was that even though technology can amplify victims’ voices, it is often used to silence them.

“The same social dynamics that keep people from disclosing their violent experiences to formal reporting sources offline also happen online,” she said.

Bringing the Cardiff Model to the U.S.

Before arriving at Tech, Foriest worked for eight years as an injury prevention coordinator at Grady Memorial Hospital in Atlanta. She implemented a trauma recovery center and Atlanta’s first hospital-based violence intervention program.

While in that position, she worked with the Cardiff Model for Violence Prevention, a public health approach to violence prevention developed by researchers at Cardiff University in Wales.

The Cardiff model’s philosophy is that violence prevention is best achieved when the healthcare and law enforcement sectors combine geographical data to determine where violence occurs in a community. 

“The Cardiff model taught Wales there was a lot about violence they didn’t know from police data alone,” Foriest said. 

One example is that researchers learned an alarming number of hospital patients were brought in from local taverns. This finding informed policymakers to implement new regulations, such as changing licensing requirements and serving alcohol in toughened glasses or non-glass vessels so they can’t be used as weapons.

In 2011, the city of Cardiff reported a 42% reduction in hospital admissions for hospital injuries. It wasn’t long before the researchers in the U.S. began importing the Cardiff model. In 2018, it became an official policy of the Centers for Disease Control and Prevention (CDC). 

The U.S. Department of Justice found in 2022 that 58% of violent crimes were not reported to law enforcement. Sixteen cities that make up the Cardiff Model for Violence Prevention National Networkare now gathering and mapping patient-reported violent injury data from hospitals to fill that data gap. 

Atlanta is one of the cities in that network, and Foriest has been an on-the-ground researcher collecting that data. Her work with the Cardiff model seamlessly integrated into her Ph.D. research as she sought ways to turn technology into a safe avenue of violence disclosure.

Working with Alex Godwin, a former Ph.D. student at Georgia Tech who is now an assistant professor at American University, she helped develop a user interface and mapping algorithm. The tool allows hospital patients who are violence victims to identify the location of the violent incident they experienced.

Foriest said, “Around the Covid-19 pandemic, we had challenges getting patients screened, and we thought we should explore different options.

“Our interface allows patients to tap down to the degree they’re comfortable on the geographic location where they were injured.

“It improved our ability to map data tremendously and decreased some of the risks patients face when disclosing violence.”

Foriest and Godwin's paper on the development of the interface tool earned an honorable mention for best paper at the 2025 Conference on Human Factors in Computing Systems (CHI) in Yokohama, Japan.

Foriest also co-authored an award-winning paper at the 2024 Conference on Computer-Supported Cooperative Work (CSCW). That paper examined how social media often silences violence victims.

Foriest is also a fellow for Data Science and Innovation at the CDC, where she continues her work on the Cardiff model. She also examines how news media coverage of suicides can often reinforce stigmas about the causes of suicide in that role.

Thriving at Tech

Foriest is entering her fifth year as a Ph.D. student, but before she came to Tech, she had no computing experience. She applied to numerous Ph.D. programs but was eventually persuaded that technology could complement her public health expertise and her goal of preventing violence.

“Tech was the only place where I could gain a new skill set while doing the things that I wanted to do in research,” she said. 

“That felt like the best fit for me, where I would get the most out of my training. I was encouraged by faculty and my peers to recognize that my perspective is valuable, and I can speak from that place and bridge my knowledge with HCI concepts.”

The crowd grew so large that no one could move in the alley, resulting in the deadliest crowd crush in the nation’s history. Nearly 160 people were killed, and another 196 were injured.

Soonho Kwon, a first-year human-centered computing Ph.D. student at Georgia Tech, lived within walking distance of the alley when the incident occurred.

“It was tragic,” Kwon said. “It really makes you think about how life is fragile. Everyone in my community talked about what it would have been like if they were in that alleyway.”

Many of the victims were young people — some of them teens who had no identification on them. Kwon thought about their family members being told their loved ones’ lives had been cut short. He wondered what memories those families would have of the deceased.

The incident inspired Kwon to create a new mobile platform that helps young adults and career professionals create a post-death memorial for their families. The platform, which Kwon and his research collaborators named Timeless, allows users to be remembered how they want to be remembered in the event of their untimely death.

“Most death preparation services are for terminally ill patients or aging adults, focusing on will management or funeral planning,” Kwon said. “We thought such needs may differ for young adults and asked how we could design a system that better caters to their needs.”

Timeless is a photo-based death preparation system that enables users to send a physical package containing pre-curated pictures, voice recordings, and letters to a designated recipient in the event of their passing. 

The system syncs with a user’s mobile photo album and creates a list of scanned faces. Users can select a face and view all the photos they’ve taken with that person. They can choose which photos they want sent to that person after death and write individual messages for each image.

Once the user’s death has been reported, Timeless sends a package to each selected individual with printed photos, letters, and a QR code or a CD that contains videos or voice recordings.

Breaking the Ice

Kwon and his collaborators designed Timeless based on a group study that asked participants to imagine what would happen if they unexpectedly died. The participants were asked what was on their bucket lists, their epitaphs, and what they would wish for if they could make one wish come true.

“Surprisingly, people were happy to participate because we framed it in a way that wasn’t gloomy,” Kwon said. “Many shared that reflecting on their death motivated them to actively express their love and be grateful for what they have. Turning something as heavy as death into something positive was a key design implication.”

Digital vs. Physical

Kwon began his research career examining virtual commemoration systems, including Facebook and Instagram commemoration pages, during the Covid-19 pandemic and exploring how technology can meaningfully memorialize the deceased.

He said two aspects distinguish Timeless from other commemoration platforms: 

• The deceased can decide how and by whom they want to be remembered.
• The fusion of digital memorialization with physical memorialization

“Leveraging only the digital side of it can be superficial,” Kwon said. “We build monuments, statues, and tombstones because the notion of death itself is losing your physical presence. By making it physical, we aimed to connect the discussion on digital legacies to traditional human commemoration forms.”

AI Afterlife

Kwon also said he is aware of artificial intelligence (AI) afterlife. This emerging technology allows people to train an AI agent and produce digital avatars with which family and friends can communicate after they die.

Meredith Ringel Morris, director and principal scientist for human-AI interaction at Google DeepMind, spoke about AI afterlife in October during the Summit on AI, Responsible Computing, and Society hosted by Georgia Tech’s School of Interactive Computing.

In her talk, Morris spoke about the criticism AI afterlife is already facing for causing people to experience extended grief and the inability to move on from losing a loved one.

Kwon said another drawback is that AI agents are susceptible to hallucinations and could say untrue things about the deceased. 

“How can you say for sure that the representation of AI is me?” he said. “As researchers, our role is to explore and critically examine how the emergence of such technology may shape society while striving to ensure its development benefits people.” 

Kwon sees Timeless as a catalyst for meaningful discussions about how a digital legacy curation system may accurately reflect a user’s wishes before death. 

He will present a paper on Timeless's design process and its implications at the 2025 ACM Conference on Human Factors in Computing Systems (CHI) this week in Yokohama, Japan.

Ph.D. candidate Md Abdur Rahaman’s dissertation studies brain data to understand how changes in brain activity shape behavior. 

Computational tools Rahaman developed for his dissertation look for informative patterns between the brain and behavior. Successful tests of his algorithms show promise to help doctors diagnose mental health disorders and design individualized treatment plans for patients.

“I've always been fascinated by the human brain and how it defines who we are,” Rahaman said. 

“The fact that so many people silently suffer from neuropsychiatric disorders, while our understanding of the brain remains limited, inspired me to develop tools that bring greater clarity to this complexity and offer hope through more compassionate, data-driven care.”

Rahaman’s dissertation introduces a framework focusing on granular factoring. This computing technique stratifies brain data into smaller, localized subgroups, making it easier for computers and researchers to study data and find meaningful patterns.

Granular factoring overcomes the challenges of size and heterogeneity in neurological data science. Brain data is obtained from neuroimaging, genomics, behavioral datasets, and other sources. The large size of each source makes it a challenge to study them individually, let alone analyze them simultaneously, to find hidden inferences. 

Rahaman’s research allows researchers and physicians to move past one-size-fits-all approaches. Instead of manually reviewing tests and scans, algorithms look for patterns and biomarkers in the subgroups that otherwise go undetected, especially ones that indicate neuropsychiatric disorders.

“My dissertation advances the frontiers of computational neuroscience by introducing scalable and interpretable models that navigate brain heterogeneity to reveal how neural dynamics shape behavior,” Rahaman said. 

“By uncovering subgroup-specific patterns, this work opens new directions for understanding brain function and enables more precise, personalized approaches to mental health care.”

Rahaman defended his dissertation on April 14, the final step in completing his Ph.D. in computational science and engineering. He will graduate on May 1 at Georgia Tech’s Ph.D. Commencement. 

After walking across the stage at McCamish Pavilion, Rahaman’s next step in his career is to go to Amazon, where he will work in the generative artificial intelligence (AI) field. 

Graduating from Georgia Tech is the summit of an educational trek spanning over a decade. Rahaman hails from Bangladesh where he graduated from Chittagong University of Engineering and Technology in 2013. He attained his master’s from the University of New Mexico in 2019 before starting at Georgia Tech. 

“Munna is an amazingly creative researcher,” said Vince Calhoun, Rahman’s advisor. Calhoun is the founding director of the Translational Research in Neuroimaging and Data Science Center (TReNDS).

TReNDS is a tri-institutional center spanning Georgia Tech, Georgia State University, and Emory University that develops analytic approaches and neuroinformatic tools. The center aims to translate the approaches into biomarkers that address areas of brain health and disease.    

“His work is moving the needle in our ability to leverage multiple sources of complex biological data to improve understanding of neuropsychiatric disorders that have a huge impact on an individual’s livelihood,” said Calhoun.

Gianna Fiduccia didn’t know much about Georgia Tech when deciding where to go to college, but with an affinity for STEM and two generations of chemical engineers in her family, she instantly felt a connection to the Institute during her first visit in 2021.   

Second place was awarded to Team Sensible, made up of juniors Oluwatooni Alade, computer science; Brandon Parker, computer science; Angela Duodu, computer science; Jesus Sierra Jr., computer science; and Hadley Williams, computer engineering. Sensible is a browser extension that rates the sustainability of products users find online and offers alternative products for items that score low. 

Third place went to Team Onyc, which includes Yasmine Green, a first-year mechanical engineering student. Onyc replaces the computer mouse with a wearable alternative that allows users to control computer navigation with the movement of their fingertips and fingernails.

Dozens of teams competed at the showcase, which is the culmination of I2P, a CREATE-X course focused on supporting students in creating solutions. The course offers research credit (for undergraduates only), up to $500 in reimbursements for physical material expenses, the opportunity to work collaboratively across majors, and faculty mentorship. It is held in the spring, summer, and fall, and it’s open to undergraduate and graduate students from all majors.  

Read our Q&A with the winner and stay tuned for our interviews with the other winning teams.

Team Carchive

Jack Rose, Junior, Computer Science

Why did you pursue your startup? 

Rose: I’ve been into cars my whole life. Trying to track cars is my hobby. There are always edge cases, and how are you planning to attack them? Because I spent so much time, especially working with other people, getting this data, and trying to figure this out, I became very adept at understanding the data. The dealers, collectors especially, were trying to understand the whole story, so they would come to me. But the way I had to do it was spreadsheets all over the place, and I was trying to find a solution to keep it all in one spot. I couldn’t find a way to do it, so I said, “Well, I’ll build it.” And then I got into I2P.

What was challenging about building your prototype over the semester? 

Rose: This semester, it was mainly trying to come up with the schema and how to physically account for the edge cases. It’s not easy; it took a lot of deep thought, discussions with other people who are into these niche cars, and understanding what details we needed. I’m still trying to add more things and figure it out. It’s not perfect, but it’s enough.

What was your favorite part about I2P? 

Rose: Adding features that I was looking for. For example, let’s say I was looking for a car. Filter all the cars over 25 years old and imported to the U.S. — I can easily search my database.

What would you say to students who are interested in entrepreneurship? 

Rose: It’s always, “You should have started sooner.” I’ve always thought about it. My biggest advice is to just start doing it, even if it’s a little bit here, a little bit there. If it doesn’t work out, at least you’ve tried.

A photo gallery from the Spring 2025 I2P Showcase can be viewed on the CREATE-X Flickr page.

Students interested in the I2P program can register for the upcoming summer and fall semesters. The deadline for Summer 2025 is May 14, and the deadline for Fall 2025 is May 16.

CREATE-X's next event, Demo Day, will take place on Aug. 28 at Exhibition Hall, where more than 100 startups will be on display. Attendees can experience the newest batch of founders leveraging the latest technology to solve pressing challenges. The event offers an opportunity to network with entrepreneurs, industry leaders, and passionate enthusiasts, and supports the next generation of innovators. Register for Demo Day today and be a part of these founders’ journeys!  

Twenty Georgia Tech students have been awarded Graduate Research Fellowships (GRF) from the National Science Foundation. The fellowships — valued at $159,000 — include funding for three years of graduate study and tuition. Graduate students pursuing full-time, research-based master’s and doctoral degrees in science, technology, engineering, and mathematics (STEM) or STEM education are eligible for the fellowship. 

Wang is a recipient of the 2025 Outstanding Dissertation Award from the Association for Computing Machinery Special Interest Group on Computer-Human Interaction (ACM SIGCHI). The award recognizes Wang for his lifelong work on democratizing human-centered AI.

“Throughout my Ph.D. and industry internships, I observed a gap in existing research: there is a strong need for practical tools for applying human-centered approaches when designing AI systems,” said Wang, now a safety researcher at OpenAI.

“My work not only helps people understand AI and guide its behavior but also provides user-friendly tools that fit into existing workflows.”

[Related: Georgia Tech College of Computing Swarms to Yokohama, Japan, for CHI 2025]

Wang’s dissertation presented techniques in visual explanation and interactive guidance to align AI models with user knowledge and values. The work culminated from years of research, fellowship support, and internships.

Wang’s most influential projects formed the core of his dissertation. These included:

• CNN Explainer: an open-source tool developed for deep-learning beginners. Since its release in July 2020, more than 436,000 global visitors have used the tool.
• DiffusionDB: a first-of-its-kind large-scale dataset that lays a foundation to help people better understand generative AI. This work could lead to new research in detecting deepfakes and designing human-AI interaction tools to help people more easily use these models.
• GAM Changer: an interface that empowers users in healthcare, finance, or other domains to edit ML models to include knowledge and values specific to their domain, which improves reliability.
• GAM Coach: an interactive ML tool that could help people who have been rejected for a loan by automatically letting an applicant know what is needed for them to receive loan approval.
• Farsight: a tool that alerts developers when they write prompts in large language models that could be harmful and misused.  

“I feel extremely honored and lucky to receive this award, and I am deeply grateful to many who have supported me along the way, including Polo, mentors, collaborators, and friends,” said Wang, who was advised by School of Computational Science and Engineering (CSE) Professor Polo Chau.

“This recognition also inspired me to continue striving to design and develop easy-to-use tools that help everyone to easily interact with AI systems.”

Like Wang, Chau advised Georgia Tech alumnus Fred Hohman (Ph.D. CSE 2020). Hohman won the ACM SIGCHI Outstanding Dissertation Award in 2022.

Chau’s group synthesizes machine learning (ML) and visualization techniques into scalable, interactive, and trustworthy tools. These tools increase understanding and interaction with large-scale data and ML models. 

Chau is the associate director of corporate relations for the Machine Learning Center at Georgia Tech. Wang called the School of CSE his home unit while a student in the ML program under Chau.

Wang is one of five recipients of this year’s award to be presented at the 2025 Conference on Human Factors in Computing Systems (CHI 2025). The conference occurs April 25-May 1 in Yokohama, Japan. 

SIGCHI is the world’s largest association of human-computer interaction professionals and practitioners. The group sponsors or co-sponsors 26 conferences, including CHI.

Wang’s outstanding dissertation award is the latest recognition of a career decorated with achievement.

Months after graduating from Georgia Tech, Forbes named Wang to its 30 Under 30 in Science for 2025 for his dissertation. Wang was one of 15 Yellow Jackets included in nine different 30 Under 30 lists and the only Georgia Tech-affiliated individual on the 30 Under 30 in Science list.

While a Georgia Tech student, Wang earned recognition from big names in business and technology. He received the Apple Scholars in AI/ML Ph.D. Fellowship in 2023 and was in the 2022 cohort of the J.P. Morgan AI Ph.D. Fellowships Program.

Along with the CHI award, Wang’s dissertation earned him awards this year at banquets across campus. The Georgia Tech chapter of Sigma Xi presented Wang with the Best Ph.D. Thesis Award. He also received the College of Computing’s Outstanding Dissertation Award.

“Georgia Tech attracts many great minds, and I’m glad that some, like Jay, chose to join our group,” Chau said. “It has been a joy to work alongside them and witness the many wonderful things they have accomplished, and with many more to come in their careers.”

Ph.D. student Phillip Si and Assistant Professor Peng Chen developed Latent-EnSF, a technique that improves how ML models assimilate data to make predictions.

In experiments predicting medium-range weather forecasting and shallow water wave propagation, Latent-EnSF demonstrated higher accuracy, faster convergence, and greater efficiency than existing methods for sparse data assimilation.

“We are currently involved in an NSF-funded project aimed at providing real-time information on extreme flooding events in Pinellas County, Florida,” said Si, who studies computational science and engineering (CSE). 

“We're actively working on integrating Latent-EnSF into the system, which will facilitate accurate and synchronized modeling of natural disasters. This initiative aims to enhance community preparedness and safety measures in response to flooding risks.” 

Latent-EnSF outperformed three comparable models in assimilation speed, accuracy, and efficiency in shallow water wave propagation experiments. These tests show models can make better and faster predictions of coastal flood waves, tides, and tsunamis. 

In experiments on medium-range weather forecasting, Latent-EnSF surpassed the same three control models in accuracy, convergence, and time. Additionally, this test demonstrated Latent-EnSF's scalability compared to other methods.

These promising results support using ML models to simulate climate, weather, and other complex systems.

Traditionally, such studies require employment of large, energy-intensive supercomputers. However, advances like Latent-EnSF are making smaller, more efficient ML models feasible for these purposes.

The Georgia Tech team mentioned this comparison in its paper. It takes hours for the European Center for Medium-Range Weather Forecasts computer to run its simulations. Conversely, the ML model FourCastNet calculated the same forecast in seconds.

“Resolution, complexity, and data-diversity will continue to increase into the future,” said Chen, an assistant professor in the School of CSE. 

“To keep pace with this trend, we believe that ML models and ML-based data assimilation methods will become indispensable for studying large-scale complex systems.”

Data assimilation is the process by which models continuously ingest new, real-world data to update predictions. This data is often sparse, meaning it is limited, incomplete, or unevenly distributed over time. 

Latent-EnSF builds on the Ensemble Filter Scores (EnSF) model developed by Florida State University and Oak Ridge National Laboratory researchers. 

EnSF’s strength is that it assimilates data with many features and unpredictable relationships between data points. However, integrating sparse data leads to lost information and knowledge gaps in the model. Also, such large models may stop learning entirely from small amounts of sparse data.

The Georgia Tech researchers employ two variational autoencoders (VAEs) in Latent-EnSF to help ML models integrate and use real-world data. The VAEs encode sparse data and predictive models together in the same space to assimilate data more accurately and efficiently.

Integrating models with new methods, like Latent-EnSF, accelerates data assimilation. Producing accurate predictions more quickly during real-world crises could save lives and property for communities.

[Related: University of South Florida Researchers Track Flooding in Coastal Communities During Hurricanes Helene and Milton]

To share Latent-EnSF to the broader research community, Chen and Si presented their paper at the SIAM Conference on Computational Science and Engineering (CSE25). The Society of Industrial and Applied Mathematics (SIAM) organized CSE25, held March 3-7 in Fort Worth, Texas.

Chen was one of ten School of CSE faculty members who presented research at CSE25, representing one-third of the School’s faculty body. Latent-EnSF was one of 15 papers by School of CSE authors and one of 23 Georgia Tech papers presented at the conference.

The pair will also present Latent-EnSF at the upcoming International Conference on Learning Representations (ICLR 2025). Occurring April 24-28 in Singapore, ICLR is one of the world’s most prestigious conferences dedicated to artificial intelligence research.

“We hope to bring attention to experts and domain scientists the exciting area of ML-based data assimilation by presenting our paper,” Chen said. “Our work offers a new solution to address some of the key shortcomings in the area for broader applications.”

Device prototypes like these usually require massive amounts of time and external resources to build, but thanks to the Uncommon Sense Lab, they can now be conveniently developed on Georgia Tech’s campus.

The lab is housed in Georgia Tech’s School of Interactive Computing and is managed by Assistant Professor Alexander Adams.

“Our overall goal is to give better access to healthcare,” Adams said. “We’re always looking at who we’re doing this for, how we’re getting it to them, how it addresses specific needs, and how to make it as financially accessible as possible.

“There’s always a space for high-end, high-precision equipment, but not everyone has access, and people are often afraid to get checked out because of the cost. If we can build something that doesn’t necessarily give someone a perfect measurement of a condition, but it can tell them they should go to the doctor, that might be enough to save a life.”

The lab provides resources to interdisciplinary researchers with backgrounds in computing, robotics, mechanical engineering, electrical engineering, and biomedical engineering to develop novel sensing and feedback system prototypes.

“We render physical prototypes that would be difficult to build without a centralized location for these resources,” said Adams, who is affiliated with the Institute for Robotics and Intelligent Machines and the Parker H. Petit Institute for Bioengineering and Bioscience. “We give students access to the tools and knowledge to build things that would typically seem unreachable.

“There’s nowhere else on campus with this collective that can go end-to-end from mechanical engineering to biomedical engineering to electrical engineering to usability.”

Examples of current prototypes being developed in the lab include a device that trains people with post-traumatic stress disorder to breathe in more regular patterns, and another that measures a person’s heart rate when they vape.

“We want to learn more about that behavior through these sensing devices, and then we’ll look at figuring out how we can help people correct their breathing patterns or quit their addiction,” Adams said.

The Uncommon Sense Lab offers high-tech, state-of-the-art machinery, including:

• 3D printers, including fused deposition modeling (FDM) printers for multi-material, high-precision prints
• A laser cutter for producing printed circuit boards (PCBs)
• Surface mount PCB manufacturing station with soldering tools, paste dispensers, and rework stations
• Optical work benches for optical system design, including microscopes and fluidics workstations
• Resin materials for casting and molding prosthetics
• Vacuum chambers and pressure chambers
• Saws, mills, lathes, and other mechanical tools for processing wood and soft metals
• Saws, grinders, polishers, and other wet tools for glass, stone, and ceramics

Since he started at the School of Interactive Computing in 2022, Adams has envisioned the lab. The lab space in the Technology Square Research Building in Midtown was thoroughly renovated, including access control, a new ceiling grid, environmental controls, pressurized air, plumbing, and vacuum and air filtration systems.

“This is the result of having built two labs at previous institutions, what I’ve learned about my type of work and my field, and what the most useful things are to handle our diverse projects,” he said.

“One of the reasons I came to Georgia Tech was because they saw the value of being interdisciplinary in a computing world and having a full lab space instead of just an office.” 

Adams said the lab will accelerate the timelines of current projects for the researchers who use it and create more bandwidth for them to take on more projects.

“I want my students to have everything at hand instead of waiting every time we need to do something,” he said. “This space is for someone who might have an idea for a remote diagnostic tool, but they’re wondering how to build it, add computation, and test it. This is the solution for those wondering how they can do that without spending a year finding and organizing access to facilities or ordering various parts.”

Adams said the lab is not a public space, but anyone interested in using it can make a written request for access. The work must be part of a collaboration, and faculty must provide funds to use resources. Access is contingent upon passing several safety courses and in-person training.

Students and faculty from the School of Computational Science and Engineering (CSE) are leading the Georgia Tech contingent at the SIAM Conference on Computational Science and Engineering (CSE25). The Society of Industrial and Applied Mathematics (SIAM) organizes CSE25, occurring March 3-7 in Fort Worth, Texas.

At CSE25, the School of CSE researchers are presenting papers that apply computing approaches to varying fields, including:                   

• Experiment designs to accelerate the discovery of material properties
• Machine learning approaches to model and predict weather forecasting and coastal flooding
• Virtual models that replicate subsurface geological formations used to store captured carbon dioxide
• Optimizing systems for imaging and optical chemistry
• Plasma physics during nuclear fusion reactions

[Related: GT CSE at SIAM CSE25 Interactive Graphic] 

“In CSE, researchers from different disciplines work together to develop new computational methods that we could not have developed alone,” said School of CSE Professor Edmond Chow. 

“These methods enable new science and engineering to be performed using computation.” 

CSE is a discipline dedicated to advancing computational techniques to study and analyze scientific and engineering systems. CSE complements theory and experimentation as modes of scientific discovery. 

Held every other year, CSE25 is the primary conference for the SIAM Activity Group on Computational Science and Engineering (SIAG CSE). School of CSE faculty serve in key roles in leading the group and preparing for the conference.

In December, SIAG CSE members elected Chow to a two-year term as the group’s vice chair. This election comes after Chow completed a term as the SIAG CSE program director. 

School of CSE Associate Professor Elizabeth Cherry has co-chaired the CSE25 organizing committee since the last conference in 2023. Later that year, SIAM members reelected Cherry to a second, three-year term as a council member at large. 

At Georgia Tech, Chow serves as the associate chair of the School of CSE. Cherry, who recently became the associate dean for graduate education of the College of Computing, continues as the director of CSE programs. 

“With our strong emphasis on developing and applying computational tools and techniques to solve real-world problems, researchers in the School of CSE are well positioned to serve as leaders in computational science and engineering both within Georgia Tech and in the broader professional community,” Cherry said. 

Georgia Tech’s School of CSE was first organized as a division in 2005, becoming one of the world’s first academic departments devoted to the discipline. The division reorganized as a school in 2010 after establishing the flagship CSE Ph.D. and M.S. programs, hiring nine faculty members, and attaining substantial research funding.

Ten School of CSE faculty members are presenting research at CSE25, representing one-third of the School’s faculty body. Of the 23 accepted papers written by Georgia Tech researchers, 15 originate from School of CSE authors.

The list of School of CSE researchers, paper titles, and abstracts includes:
Bayesian Optimal Design Accelerates Discovery of Material Properties from Bubble Dynamics
Postdoctoral Fellow Tianyi Chu, Joseph Beckett, Bachir Abeid, and Jonathan Estrada (University of Michigan), Assistant Professor Spencer Bryngelson
[Abstract]

Latent-EnSF: A Latent Ensemble Score Filter for High-Dimensional Data Assimilation with Sparse Observation Data
Ph.D. student Phillip Si, Assistant Professor Peng Chen
[Abstract]

A Goal-Oriented Quadratic Latent Dynamic Network Surrogate Model for Parameterized Systems
Yuhang Li, Stefan Henneking, Omar Ghattas (University of Texas at Austin), Assistant Professor Peng Chen
[Abstract]

Posterior Covariance Structures in Gaussian Processes
Yuanzhe Xi (Emory University), Difeng Cai (Southern Methodist University), Professor Edmond Chow
[Abstract]

Robust Digital Twin for Geological Carbon Storage
Professor Felix Herrmann, Ph.D. student Abhinav Gahlot, alumnus Rafael Orozco (Ph.D. CSE-CSE 2024), alumnus Ziyi (Francis) Yin (Ph.D. CSE-CSE 2024), and Ph.D. candidate Grant Bruer
[Abstract]

Industry-Scale Uncertainty-Aware Full Waveform Inference with Generative Models
Rafael Orozco, Ph.D. student Tuna Erdinc, alumnus Mathias Louboutin (Ph.D. CS-CSE 2020), and Professor Felix Herrmann
[Abstract]

Optimizing Coupled Systems: Insights from Co-Design Imaging and Optical Chemistry
Assistant Professor Raphaël Pestourie, Wenchao Ma and Steven Johnson (MIT), Lu Lu (Yale University), Zin Lin (Virginia Tech)
[Abstract]

Multifidelity Linear Regression for Scientific Machine Learning from Scarce Data
Assistant Professor Elizabeth Qian, Ph.D. student Dayoung Kang, Vignesh Sella, Anirban Chaudhuri and Anirban Chaudhuri (University of Texas at Austin)
[Abstract]

LyapInf: Data-Driven Estimation of Stability Guarantees for Nonlinear Dynamical Systems
Ph.D. candidate Tomoki Koike and Assistant Professor Elizabeth Qian
[Abstract]

The Information Geometric Regularization of the Euler Equation
Alumnus Ruijia Cao (B.S. CS 2024), Assistant Professor Florian Schäfer
[Abstract]

Maximum Likelihood Discretization of the Transport Equation
Ph.D. student Brook Eyob, Assistant Professor Florian Schäfer
[Abstract]

Intelligent Attractors for Singularly Perturbed Dynamical Systems
Daniel A. Serino (Los Alamos National Laboratory), Allen Alvarez Loya (University of Colorado Boulder), Joshua W. Burby, Ioannis G. Kevrekidis (Johns Hopkins University), Assistant Professor Qi Tang (Session Co-Organizer)
[Abstract]

Accurate Discretizations and Efficient AMG Solvers for Extremely Anisotropic Diffusion Via Hyperbolic Operators
Golo Wimmer, Ben Southworth, Xianzhu Tang (LANL), Assistant Professor Qi Tang 
[Abstract]

Randomized Linear Algebra for Problems in Graph Analytics
Professor Rich Vuduc
[Abstract]

Improving Spgemm Performance Through Reordering and Cluster-Wise Computation
Assistant Professor Helen Xu
[Abstract]

While this has stimulated economic growth, especially in Medellín, Antioquia’s capital and largest city, it has also introduced challenges such as rising costs and gentrification, with certain neighborhoods becoming less affordable.

Comfama is a nonprofit organization that strives to grow the middle class by providing social and economic services to families. The organization has begun a groundbreaking project in collaboration with the Business Analytics Center (BAC) at the Georgia Tech Scheller College of Business to enhance its ability to forecast and meet the needs of the population.

Determining the Future of Compensation Funds

Comfama is a “compensation fund.” In Colombia, these private, nonprofit organizations have been created to improve the lives of workers and their families. They provide social services for recreation, culture, education, preventive healthcare, housing assistance, loans, and more. There are 42 compensation funds across the country that play a vital part in the country's social security system, according to Santiago García Rivera, head of the Information and Analytics Laboratory at Comfama.

For companies in Colombia, participation in a compensation fund is mandatory. Each fund collects a 4 percent payroll tax from affiliated companies to provide monetary subsidies to workers and their families. "We have about 121 thousand companies affiliated with Comfama, which includes about 1.4 million workers. When you take into account their families, that's around 2.7 million people we serve, plus a large group of non affiliated people that use our services," said García Rivera.

For Comfama, economic and demographic shifts have complicated the prediction of how many people will use its services. "Recognizing these challenges, Comfama is embracing data-driven solutions. We want to build a robust prediction model to help us forecast what will happen to our affiliated population in the future," he said.

The Georgia Tech Connection

This is where Georgia Tech comes in. One of García Rivera's colleagues at Comfama, Juan David Penagos, head of Ventures and New Business Development, knew about the Georgia Tech Enterprise Innovation Institute Medellín Innovation and Entrepreneurship Center and suggested they reach out to see whether they could put a project together with business analytics students. Sara Araujo Santos, managing director of Development for Latin America for the Center, contacted Sherri Von Behren, the BAC's corporate engagement manager, about possibly creating an MBA business analytics practicum project to help Comfama.

Von Behren contacted Jonathan Fan, a faculty member of the Information Technology Management group in the Scheller College. Fan leads students in transforming data into business solutions through the Business Analytics Practicum course, which is offered in the fall for graduate students and in the spring for undergraduates.  Fan immediately saw the value of the opportunity for his MBA students and set up a practicum in which they are developing predictive models using time series data and macroeconomic variables.

There are two teams assigned to the project: Team Data Paisa Squad with Aaron Payne, MBA ‘26 (team lead), Lissette Chavez, MBA ‘25, and Boris Taganov, MBA ‘25, and Team The Growth Gurus of Antioquia, with Justin Siegel, MS in Analytics ’25 (team lead), Srinjoy DasMahapatra, MBA ’25, and Vinaya Vinigalla, MBA ’24. Haofei Qin, Ph.D. candiate at Scheller helped mentor students along the way. 

They meet weekly with the Comfama team, which includes analysts and data scientists Alejandra Bernal, Susanna Londoño, and Wbeimar Ossa. The teams discuss their progress and address any challenges they face that week. With less than two months to go, they're seeing results.

Fan has been pleased with the results so far. “This cohort was truly outstanding,” he said, speaking of his students. ”They approached complex topics with clarity and creativity, and their collaborative spirit led to innovative ideas and enlightening discussions. They handled challenging subjects effortlessly, always coming up with fresh and interesting perspectives."

Managing the Present and Predicting the Future

Regarding the work of Team Data Paisa Squad, Aaron Payne remarked, "One of our key successes has been developing a model framework that provides accurate forecasts and adapts to changes in external economic indicators. By integrating machine learning techniques alongside traditional time series models, we've increased our forecasts' robustness. Additionally, the collaboration with Comfama's internal team has been highly productive, enabling us to align our technical solutions with their business needs. The early feedback on our findings has been positive."

Payne stated that one of the main challenges they've faced has been combining data from multiple sources, as each source has different levels of detail and accuracy. Ensuring these data sets are consistent and reliable has been difficult, especially for economic factors that may not directly match Comfama's internal data. Another challenge is adapting standard forecasting models like SARIMAX—Seasonal AutoRegressive Integrated Moving Average with eXogenous variables— to account for external influences like government policies or unexpected economic changes.

"This experience has reinforced the value of experiential learning in advancing my business analytics skills. Working with real-world data, especially in a dynamic organization like Comfama, has provided a deeper understanding of how to apply advanced analytical methods to solve practical business problems. The practicum has helped bridge the gap between theory and practice, giving me confidence in using these tools to drive decision-making in real business environments," said Payne.

The practicum is more than just about numbers, though. It's about understanding the lives behind the data points. As Fan reminds his students, "A model is just a model, but those data points represent individual lives. We want to understand the mechanism or the story behind the data."

Researchers from the School of Computational Science and Engineering (CSE) created the Large Pre-Trained Time-Series Model (LPTM) framework. LPTM is a single foundational model that completes forecasting tasks across a broad range of domains. 

Along with performing as well or better than models purpose-built for their applications, LPTM requires 40% less data and 50% less training time than current baselines. In some cases, LPTM can be deployed without any training data.

The key to LPTM is that it is pre-trained on datasets from different industries like healthcare, transportation, and energy. The Georgia Tech group created an adaptive segmentation module to make effective use of these vastly different datasets.

The Georgia Tech researchers will present LPTM in Vancouver, British Columbia, Canada, at the 2024 Conference on Neural Information Processing Systems (NeurIPS 2024). NeurIPS is one of the world’s most prestigious conferences on artificial intelligence (AI) and ML research.

“The foundational model paradigm started with text and image, but people haven’t explored time-series tasks yet because those were considered too diverse across domains,” said B. Aditya Prakash, one of LPTM’s developers. 

“Our work is a pioneer in this new area of exploration where only few attempts have been made so far.”

[MICROSITE: Georgia Tech at NeurIPS 2024]

Foundational models are trained with data from different fields, making them powerful tools when assigned tasks. Foundational models drive GPT, DALL-E, and other popular generative AI platforms used today. LPTM is different though because it is geared toward time-series, not text and image generation.  

The Georgia Tech researchers trained LPTM on data ranging from epidemics, macroeconomics, power consumption, traffic and transportation, stock markets, and human motion and behavioral datasets.

After training, the group pitted LPTM against 17 other models to make forecasts as close to nine real-case benchmarks. LPTM performed the best on five datasets and placed second on the other four.

The nine benchmarks contained data from real-world collections. These included the spread of influenza in the U.S. and Japan, electricity, traffic, and taxi demand in New York, and financial markets.   

The competitor models were purpose-built for their fields. While each model performed well on one or two benchmarks closest to its designed purpose, the models ranked in the middle or bottom on others.

In another experiment, the Georgia Tech group tested LPTM against seven baseline models on the same nine benchmarks in zero-shot forecasting tasks. Zero-shot means the model is used out of the box and not given any specific guidance during training. LPTM outperformed every model across all benchmarks in this trial.

LPTM performed consistently as a top-runner on all nine benchmarks, demonstrating the model’s potential to achieve superior forecasting results across multiple applications with less and resources.

“Our model also goes beyond forecasting and helps accomplish other tasks,” said Prakash, an associate professor in the School of CSE. 

“Classification is a useful time-series task that allows us to understand the nature of the time-series and label whether that time-series is something we understand or is new.”

One reason traditional models are custom-built to their purpose is that fields differ in reporting frequency and trends. 

For example, epidemic data is often reported weekly and goes through seasonal peaks with occasional outbreaks. Economic data is captured quarterly and typically remains consistent and monotone over time. 

LPTM’s adaptive segmentation module allows it to overcome these timing differences across datasets. When LPTM receives a dataset, the module breaks data into segments of different sizes. Then, it scores all possible ways to segment data and chooses the easiest segment from which to learn useful patterns.

LPTM’s performance, enhanced through the innovation of adaptive segmentation, earned the model acceptance to NeurIPS 2024 for presentation. NeurIPS is one of three primary international conferences on high-impact research in AI and ML. NeurIPS 2024 occurs Dec. 10-15.

Ph.D. student Harshavardhan Kamarthi partnered with Prakash, his advisor, on LPTM. The duo are among the 162 Georgia Tech researchers presenting over 80 papers at the conference. 

Prakash is one of 46 Georgia Tech faculty with research accepted at NeurIPS 2024. Nine School of CSE faculty members, nearly one-third of the body, are authors or co-authors of 17 papers accepted at the conference. 

Along with sharing their research at NeurIPS 2024, Prakash and Kamarthi released an open-source library of foundational time-series modules that data scientists can use in their applications.

“Given the interest in AI from all walks of life, including business, social, and research and development sectors, a lot of work has been done and thousands of strong papers are submitted to the main AI conferences,” Prakash said. 

“Acceptance of our paper speaks to the quality of the work and its potential to advance foundational methodology, and we hope to share that with a larger audience.”

SubjECTive-QA is the first human-curated dataset on question-answer pairs from earnings call transcripts (ECTs). The dataset teaches models to identify subjective features in ECTs, like clarity and cautiousness.   

The dataset lays the foundation for a new approach to identifying disinformation and misinformation caused by nuances in speech. While ECT responses can be technically true, unclear or irrelevant information can misinform stakeholders and affect their decision-making. 

Tests on White House press briefings showed that the dataset applies to other sectors with frequent question-and-answer encounters, notably politics, journalism, and sports. This increases the odds of effectively informing audiences and improving transparency across public spheres.   

The intersecting work between natural language processing and finance earned the paper acceptance to NeurIPS 2024, the 38th Annual Conference on Neural Information Processing Systems. NeurIPS is one of the world’s most prestigious conferences on artificial intelligence (AI) and machine learning (ML) research.

"SubjECTive-QA has the potential to revolutionize nowcasting predictions with enhanced clarity and relevance,” said Agam Shah, the project’s lead researcher. 

“Its nuanced analysis of qualities in executive responses, like optimism and cautiousness, deepens our understanding of economic forecasts and financial transparency."

[MICROSITE: Georgia Tech at NeurIPS 2024]

SubjECTive-QA offers a new means to evaluate financial discourse by characterizing language's subjective and multifaceted nature. This improves on traditional datasets that quantify sentiment or verify claims from financial statements.

The dataset consists of 2,747 Q&A pairs taken from 120 ECTs from companies listed on the New York Stock Exchange from 2007 to 2021. The Georgia Tech researchers annotated each response by hand based on six features for a total of 49,446 annotations.

The group evaluated answers on:

• Relevance: the speaker answered the question with appropriate details.
• Clarity: the speaker was transparent in the answer and the message conveyed.
• Optimism: the speaker answered with a positive outlook regarding future outcomes.
• Specificity: the speaker included sufficient and technical details in their answer.
• Cautiousness: the speaker answered using a conservative, risk-averse approach.
• Assertiveness: the speaker answered with certainty about the company’s events and outcomes.

The Georgia Tech group validated their dataset by training eight computer models to detect and score these six features. Test models comprised of three BERT-based pre-trained language models (PLMs), and five popular large language models (LLMs) including Llama and ChatGPT. 

All eight models scored the highest on the relevance and clarity features. This is attributed to domain-specific pretraining that enables the models to identify pertinent and understandable material.

The PLMs achieved higher scores on the clear, optimistic, specific, and cautious categories. The LLMs scored higher in assertiveness and relevance. 

In another experiment to test transferability, a PLM trained with SubjECTive-QA evaluated 65 Q&A pairs from White House press briefings and gaggles. Scores across all six features indicated models trained on the dataset could succeed in other fields outside of finance. 

"Building on these promising results, the next step for SubjECTive-QA is to enhance customer service technologies, like chatbots,” said Shah, a Ph.D. candidate studying machine learning. 

“We want to make these platforms more responsive and accurate by integrating our analysis techniques from SubjECTive-QA."

SubjECTive-QA culminated from two semesters of work through Georgia Tech’s Vertically Integrated Projects (VIP) Program. The VIP Program is an approach to higher education where undergraduate and graduate students work together on long-term project teams led by faculty. 

Undergraduate students earn academic credit and receive hands-on experience through VIP projects. The extra help advances ongoing research and gives graduate students mentorship experience.

Computer science major Huzaifa Pardawala and mathematics major Siddhant Sukhani co-led the SubjECTive-QA project with Shah. 

Fellow collaborators included Veer Kejriwal, Abhishek Pillai, Rohan Bhasin, Andrew DiBiasio, Tarun Mandapati, and Dhruv Adha. All six researchers are undergraduate students studying computer science. 

Sudheer Chava co-advises Shah and is the faculty lead of SubjECTive-QA. Chava is a professor in the Scheller College of Business and director of the M.S. in Quantitative and Computational Finance (QCF) program.

Chava is also an adjunct faculty member in the College of Computing’s School of Computational Science and Engineering (CSE).

"Leading undergraduate students through the VIP Program taught me the powerful impact of balancing freedom with guidance,” Shah said. 

“Allowing students to take the helm not only fosters their leadership skills but also enhances my own approach to mentoring, thus creating a mutually enriching educational experience.”

Presenting SubjECTive-QA at NeurIPS 2024 exposes the dataset for further use and refinement. NeurIPS is one of three primary international conferences on high-impact research in AI and ML. The conference occurs Dec. 10-15.

The SubjECTive-QA team is among the 162 Georgia Tech researchers presenting over 80 papers at NeurIPS 2024. The Georgia Tech contingent includes 46 faculty members, like Chava. These faculty represent Georgia Tech’s Colleges of Business, Computing, Engineering, and Sciences, underscoring the pertinence of AI research across domains. 

"Presenting SubjECTive-QA at prestigious venues like NeurIPS propels our research into the spotlight, drawing the attention of key players in finance and tech,” Shah said.

“The feedback we receive from this community of experts validates our approach and opens new avenues for future innovation, setting the stage for transformative applications in industry and academia.”

Rachel Lowy, a Ph.D. student in the School of Interactive Computing, piloted a new human-computer interaction design course for IDD students in Georgia Tech’s Excel program. Excel is an Inclusive Postsecondary Education (IPSE) program that offers a four-year track for IDD students to earn two separate certificates.

Lowy said the course differs from typical technology courses taught to IDD students. It provides autonomy and encourages students to contribute input on how the course is designed and which technology projects they want to create. They reflect critically on the role of technology in the world and use that reflection to design technology for themselves.

The course is also unique because it involves a mix of professional educators and technology researchers working together. Lowy taught the class alongside her advisor, Assistant Professor Jennifer Kim, her lab colleague, Kaely Hall, master’s students in the Georgia Tech MS-HCI program, computer science undergraduates, and Excel educators.

“We have a few models of students designing next to designers in classrooms, but they tend to be only taught by professionals in K-12 education, not necessarily HCI researchers in higher education. They rarely include students with IDD,” she said. 

“In higher education, HCI projects may not go further than the classroom space. This course was special because we can teach these students with IDD high-level concepts about HCI and adopt their ideas into ongoing projects. We can keep working on them after the class has finished.”

Lowy said she designed the course based on previous work on accessible co-design and consulted with Assistant Professor Jessica Roberts, an educational technology researcher in the School of IC, to develop course materials. She refined the course with her co-teachers as she taught it, responding to observations and reflections from students. 

If the students had not been allowed to provide their input, Lowy and her team would never have learned how IDD students prefer to use different technologies. Lowy said they took that feedback to implement strength-based activities. 

“So much technology design for people with disabilities focuses on what they cannot do,” she said. “Our lab likes to focus on what they can do and their strengths.”

During one class, the researchers brought a robot dog into the classroom to determine whether it could supply emotional support to the students. The feedback they received showed the students were more interested in how the robot dog could be a companion in day-to-day activities.

“We came in with an idea of how the participants might want to use the technology,” Lowy said. “The students had a much broader view of what they might like to use this technology for. They reflected on their lives, and that’s exactly what we want good design to do.”

Lowy said she hopes the course serves as a blueprint for inclusive advanced technology courses at the university level.

“Most of their technology courses focus on workplace education like how to use Microsoft Suite, Google Calendar, or Outlook,” she said. “We’re working on more of a foundational level about how those technologies are designed and whether they work for them.”

She also said the course could be a step toward more inclusiveness in university classroom environments with traditional students and students with IDD learning together.

“Something that IPSE students have told me is that it’s hard to keep up with lectures, and they sometimes struggle to keep up in class,” she said. “It’d be great if they take a class specifically targeted to them at their own pace with a hands-on element to it, and they got to learn through experiential activities. Then they take the knowledge they’ve gleaned into an inclusive class where they work with their peers.”

She also suggested other models universities might offer, like an Intro to HCI course for IDD students that allows them to work on projects with students enrolled in the traditional Intro to HCI course.

“Any university with an IPSE program and an HCI program on campus can do this,” she said.

Georgia Tech Ph.D. student Austin P. Wright is first author of a paper that introduces Nested Fusion. The new algorithm improves scientists’ ability to search for past signs of life on the Martian surface. 

In addition to supporting NASA’s Mars 2020 mission, scientists from other fields working with large, overlapping datasets can use Nested Fusion’s methods toward their studies.

Wright presented Nested Fusion at the 2024 International Conference on Knowledge Discovery and Data Mining (KDD 2024) where it was a runner-up for the best paper award. KDD is widely considered the world's most prestigious conference for knowledge discovery and data mining research.

“Nested Fusion is really useful for researchers in many different domains, not just NASA scientists,” said Wright. “The method visualizes complex datasets that can be difficult to get an overall view of during the initial exploratory stages of analysis.”

Nested Fusion combines datasets with different resolutions to produce a single, high-resolution visual distribution. Using this method, NASA scientists can more easily analyze multiple datasets from various sources at the same time. This can lead to faster studies of Mars’ surface composition to find clues of previous life.

The algorithm demonstrates how data science impacts traditional scientific fields like chemistry, biology, and geology.

Even further, Wright is developing Nested Fusion applications to model shifting climate patterns, plant and animal life, and other concepts in the earth sciences. The same method can combine overlapping datasets from satellite imagery, biomarkers, and climate data.

“Users have extended Nested Fusion and similar algorithms toward earth science contexts, which we have received very positive feedback,” said Wright, who studies machine learning (ML) at Georgia Tech.

“Cross-correlational analysis takes a long time to do and is not done in the initial stages of research when patterns appear and form new hypotheses. Nested Fusion enables people to discover these patterns much earlier.”

Wright is the data science and ML lead for PIXLISE, the software that NASA JPL scientists use to study data from the Mars Perseverance Rover.

Perseverance uses its Planetary Instrument for X-ray Lithochemistry (PIXL) to collect data on mineral composition of Mars’ surface. PIXL’s two main tools that accomplish this are its X-ray Fluorescence (XRF) Spectrometer and Multi-Context Camera (MCC).

When PIXL scans a target area, it creates two co-aligned datasets from the components. XRF collects a sample's fine-scale elemental composition. MCC produces images of a sample to gather visual and physical details like size and shape. 

A single XRF spectrum corresponds to approximately 100 MCC imaging pixels for every scan point. Each tool’s unique resolution makes mapping between overlapping data layers challenging. However, Wright and his collaborators designed Nested Fusion to overcome this hurdle.

In addition to progressing data science, Nested Fusion improves NASA scientists' workflow. Using the method, a single scientist can form an initial estimate of a sample’s mineral composition in a matter of hours. Before Nested Fusion, the same task required days of collaboration between teams of experts on each different instrument.

“I think one of the biggest lessons I have taken from this work is that it is valuable to always ground my ML and data science problems in actual, concrete use cases of our collaborators,” Wright said. 

“I learn from collaborators what parts of data analysis are important to them and the challenges they face. By understanding these issues, we can discover new ways of formalizing and framing problems in data science.”

Wright presented Nested Fusion at KDD 2024, held Aug. 25-29 in Barcelona, Spain. KDD is an official special interest group of the Association for Computing Machinery. The conference is one of the world’s leading forums for knowledge discovery and data mining research.

Nested Fusion won runner-up for the best paper in the applied data science track, which comprised of over 150 papers. Hundreds of other papers were presented at the conference’s research track, workshops, and tutorials. 

Wright’s mentors, Scott Davidoff and Polo Chau, co-authored the Nested Fusion paper. Davidoff is a principal research scientist at the NASA Jet Propulsion Laboratory. Chau is a professor at the Georgia Tech School of Computational Science and Engineering (CSE).

“I was extremely happy that this work was recognized with the best paper runner-up award,” Wright said. “This kind of applied work can sometimes be hard to find the right academic home, so finding communities that appreciate this work is very encouraging.”

Georgia Tech researchers partnered with medical experts from Children’s Healthcare of Atlanta (CHOA) to develop and test ARCollab. The iOS-based app leverages advanced AR technologies to let doctors collaborate together and interact with a patient’s 3D heart model when planning surgeries.

The usability evaluation demonstrates the app’s effectiveness, finding that ARCollab is easy to use and understand, fosters collaboration, and improves surgical planning.

“This tool is a step toward easier collaborative surgical planning. ARCollab could reduce the reliance on physical heart models, saving hours and even days of time while maintaining the collaborative nature of surgical planning,” said M.S. student Pratham Mehta, the app’s lead researcher.

“Not only can it benefit doctors when planning for surgery, it may also serve as a teaching tool to explain heart deformities and problems to patients.”

Two cardiologists and three cardiothoracic surgeons from CHOA tested ARCollab. The two-day study ended with the doctors taking a 14-question survey assessing the app’s usability. The survey also solicited general feedback and top features.

The Georgia Tech group determined from the open-ended feedback that:

• ARCollab enables new collaboration capabilities that are easy to use and facilitate surgical planning.
• Anchoring the model to a physical space is important for better interaction.
• Portability and real-time interaction are crucial for collaborative surgical planning.

Users rated each of the 14 questions on a 7-point Likert scale, with one being “strongly disagree” and seven being “strongly agree.” The 14 questions were organized into five categories: overall, multi-user, model viewing, model slicing, and saving and loading models.

The multi-user category attained the highest rating with an average of 6.65. This included a unanimous 7.0 rating that it was easy to identify who was controlling the heart model in ARCollab. The scores also showed it was easy for users to connect with devices, switch between viewing and slicing, and view other users’ interactions.

The model slicing category received the lowest, but formidable, average of 5.5. These questions assessed ease of use and understanding of finger gestures and usefulness to toggle slice direction.

Based on feedback, the researchers will explore adding support for remote collaboration. This would assist doctors in collaborating when not in a shared physical space. Another improvement is extending the save feature to support multiple states.

“The surgeons and cardiologists found it extremely beneficial for multiple people to be able to view the model and collaboratively interact with it in real-time,” Mehta said.

The user study took place in a CHOA classroom. CHOA also provided a 3D heart model for the test using anonymous medical imaging data. Georgia Tech’s Institutional Review Board (IRB) approved the study and the group collected data in accordance with Institute policies.

The five test participants regularly perform cardiovascular surgical procedures and are employed by CHOA. 

The Georgia Tech group provided each participant with an iPad Pro with the latest iOS version and the ARCollab app installed. Using commercial devices and software meets the group’s intentions to make the tool universally available and deployable.

“We plan to continue iterating ARCollab based on the feedback from the users,” Mehta said. 

“The participants suggested the addition of a ‘distance collaboration’ mode, enabling doctors to collaborate even if they are not in the same physical environment. This allows them to facilitate surgical planning sessions from home or otherwise.”

The Georgia Tech researchers are presenting ARCollab and the user study results at IEEE VIS 2024, the Institute of Electrical and Electronics Engineers (IEEE) visualization conference. 

IEEE VIS is the world’s most prestigious conference for visualization research and the second-highest rated conference for computer graphics. It takes place virtually Oct. 13-18, moved from its venue in St. Pete Beach, Florida, due to Hurricane Milton.

The ARCollab research group's presentation at IEEE VIS comes months after they shared their work at the Conference on Human Factors in Computing Systems (CHI 2024).

Undergraduate student Rahul Narayanan and alumni Harsha Karanth (M.S. CS 2024) and Haoyang (Alex) Yang (CS 2022, M.S. CS 2023) co-authored the paper with Mehta. They study under Polo Chau, a professor in the School of Computational Science and Engineering.

The Georgia Tech group partnered with Dr. Timothy Slesnick and Dr. Fawwaz Shaw from CHOA on ARCollab’s development and user testing.

"I'm grateful for these opportunities since I get to showcase the team's hard work," Mehta said.

“I can meet other like-minded researchers and students who share these interests in visualization and human-computer interaction. There is no better form of learning.”

Installed recently at Georgia Gwinnett College (GGC), an X-band weather radar purchased two years ago by the Georgia Institute of Technology and the University of Georgia (UGA) is now providing data for a section of north Georgia where information on severe storms such as tornados can be limited by terrain.

The radar will also be used for research into weather and severe storms, and by students at the three institutions for learning about everything from physics and engineering to weather, rainfall, and the effects of changing climate on the migration patterns of birds and insects. The instrument will be one of just a handful of weather radars operated by universities in the United States.

“We are really excited about this partnership with Georgia Tech, the Georgia Tech Research Institute, the University of Georgia, and Georgia Gwinnett College,” said Marshall Shepherd, Associate Dean for Research, Scholarship and Partnership at UGA’s Franklin College of Arts and Sciences and Director of UGA’s Atmospheric Sciences Program. “The radar will be a real-time component of classes, so it’s creating new instructional and service capabilities. It will also enable researchers at the University of Georgia and Georgia Tech to pursue new research opportunities in the areas of severe weather, frozen precipitation – and perhaps even studies of birds and insects.”

The radar will provide a new data source for UGA’s WeatherDawgs service, which provides hyperlocal weather data not only for the Athens community, but also for residents of eastern and northeastern Georgia. The system will also provide a real-time component for the mesoscale meteorology course taught at the university.

For Georgia Tech, the radar will support the work of the Severe Storms Research Center (SSRC), a state-funded initiative that serves as a focal point for severe storms research in the state. The radar will also support research and education at Georgia Tech, including courses on weather radar systems and studies of lightning being done in the School of Electrical and Computer Engineering.

“The new radar will help fill some low-level gaps in weather radar coverage in north Georgia, and give higher-resolution data for the Georgia Gwinnett campus, University of Georgia campus, Georgia Tech campus and areas in between,” said John Trostel, director of the SSRC. “This is an area where both UGA and Georgia Tech have interests because it goes from urban to suburban, then back to urban. We might see some very interesting weather phenomena going on in those transition areas.”

The National Weather Service has access to a feed from the radar and will use it to obtain information about low-altitude weather activity that can’t be seen as well from sources such as the NEXRAD radar based in Peachtree City and the Terminal Doppler Weather Radar at Hartsfield-Jackson Atlanta International Airport, Trostel added.

For Georgia Gwinnett College, the radar will provide real-world examples of how physics and engineering concepts are applied. Data from the radar system, which will be accessible to the college, would also provide students with a new research opportunity that is a required component of the science curriculum.

“Our Physics and Pre-Engineering courses already cover the concepts of electromagnetic waves and the Doppler effect, which are the main principles behind radar,” said Neelam Khan, the Chair of the Physics and Pre-Engineering Department at Georgia Gwinnett College. “Through this radar, students will learn about the applications of Doppler radar to track weather patterns and visualize the data it produces.”

Connections with the University of Georgia, Georgia Tech, and the Georgia Tech Research Institute will also help broaden the experience of students at Georgia Gwinnett College, a four-year public college that was founded in 2005 and now has more than 11,000 students, Khan said. All three collaborating institutions are part of the University System of Georgia.

The Furuno WR-2100 X-band weather radar was purchased in 2022 using funding from Georgia Tech and the University of Georgia. It was initially placed atop a building on GTRI’s Smyrna campus, where it underwent tests while Trostel and Shepherd searched for the best location for a more permanent installation. The researchers have used the device to look at storms, generate data, and practice data analysis.

The Georgia Gwinnett location was selected because the campus location enables coverage for both Atlanta and Athens. The Gwinnett County location also helps fill potential gaps in northeast Georgia and brings a unique resource for GGC’s educational mission. The radar is now fully operational.

Owning and operating a weather radar is unusual for colleges and universities, but not surprising given the impact of severe weather in Georgia, Shepherd noted.

“Weather is a significant threat to our lives and property, particularly in Georgia,” Shepherd said. “While we have an adequate radar network from the National Weather Service and the Terminal Doppler Weather Radar, there are often gaps and needs for higher resolution, more detailed information. Our institutions have entered very rare air in owning and operating a weather radar that will benefit our students, the state, and our research enterprise in the University System of Georgia institutions.”

Because they’ll be able to control the geographic areas covered by the radar and the level of detail in the information gathered, the new weather radar will be a useful tool not only for tracking storms, but also for conducting research, Trostel said. Its ability to provide highly detailed information even allows it to track the movement of insects and birds, for example.

“We can see things at higher resolution, and we have complete control over how we manipulate the radar beam to look at things,” Trostel said. “The radar is much less expensive to purchase and operate than other weather radars, which makes it a budget-friendly tool for university research.”

The instrument cost approximately $150,000 to purchase and was acquired through donations and internal funding at UGA and Georgia Tech. Shepherd and Tom Mote, the founding director of the Atmospheric Sciences Program at UGA, contributed funds from institutional research budgets. A significant financial gift was also acquired from Elaine Neal, an alumna of the UGA Department of Geography and longtime donor to the University of Georgia.

At Georgia Tech, funds were provided by GTRI’s Sensors and Electromagnetic Applications Laboratory, and the Aerospace, Transportation and Advanced Systems Laboratory, the Georgia Tech Office of the Executive Vice President for Research, and Georgia Tech’s College of Engineering.

Writer: John Toon (john.toon@gtri.gatech.edu)
GTRI Communications
Georgia Tech Research Institute
Atlanta, Georgia USA

The event traditionally begins shortly after the semester starts, giving the entrepreneurially curious a preview of what’s to come if they join the program’s accelerator during the next application cycle.

Demo Day is the culmination of the 12-week summer accelerator, Startup Launch, where founders receive mentorship, $5,000 in optional funding, and $150,000 in services to help build their businesses. Teams can be interdisciplinary, made up of co-founders even outside of Georgia Tech, and solopreneurs, ready to solve real-world problems.

Each year, Startup Launch has grown, from an initial cohort of eight startups to over 100 this year. The Office of Commercialization, the home of CREATE-X, plans to keep expanding opportunities for the Georgia Tech community to grow their entrepreneurial skills.

Counting courses, events, programming, and partnerships, CREATE-X has had more than 32,000 participants. The ultimate goal and mission of the program is to instill entrepreneurial confidence in all Tech students. Rahul Saxena, director of the program, spoke about how far the Institute has come in the last decade. 

“I’ve been plugged into Georgia Tech for over 10 years. In the past, when you said Georgia Tech and entrepreneurship in the same sentence, they’d laugh, believe it or not,” he said. “Fast-forward, we’re one of the top entrepreneurial schools in the country. Our first four cohorts value over $100 million, with one of them being a unicorn, and our last four cohorts are well on their way. We want our students to have as many shots at gold as possible before they graduate. And even if they decide on a traditional career pathway, we believe they’ll be ahead with this entrepreneurial mindset, which is something lacking in corporate.”

This year, CREATE-X reached over 560 startup teams launched. Founders represented 38 academic majors, and their total startup portfolio valuation exceeds $2 billion. 

CREATE-X opened its Startup Launch application for its next cohort on Aug. 30. For those interested, the priority deadline is Nov. 17. Early applicants have a higher chance at acceptance and the opportunity for more feedback. So, send in your applications to Startup Launch and become the next founder at Georgia Tech.

Missed out on Demo Day? Check out the CREATE-X Flickr page to see photos from the event and the Demo Day page to see other teams. For more opportunities to engage, visit the CREATE-X Engage page for upcoming events. 

Spotlight on Startups

Some of the standout startups from this year’s Demo Day include:

Students returning to campus for a new semester often struggle to find time for physical activities because of their new routines and schedules. However, a new app developed at Georgia Tech helps busy students prioritize physical activity in their daily routines.

Ph.D. student Kefan Xu of the Ubicomp Health and Wellness Lab at Georgia Tech created Plannergy, a time management app that identifies open time blocks in users’ schedules. 

Xu introduced Plannergy at the Conference on Human Factors in Computing (CHI) in Honolulu, Hawaii in May. He says the app is ideal for college students because they tend to have busy and inconsistent schedules.

Plannergy allows users to track their schedules, reflect on what activities would be beneficial and timely, and strategize how to implement the activity into their schedule.

"Currently, the app is catered to people who’ve been physically inactive and have inconsistent schedules,” Xu said. “College students know their schedule will change when they begin a new semester. They need to get some physical activity and find opportunities in the day they can leverage. It could be as simple as walking to school instead of taking a scooter.”

Xu tested his app on 16 college students who planned their physical activities every seven days and followed a reflective iteration framework to track improvement. The results showed that Plannergy is an effective behavior change tool. The findings also indicate that it increases participants’ awareness of their schedules.

The American Heart Association says adults can reduce the risk of heart disease by participating in at least 150 minutes of moderate-intensity physical activity weekly.

The Centers for Disease Control and Prevention released a report in 2023 that found 72% of Americans aren’t meeting that standard.

As Xu points out in his paper, studies have shown that incorporating physical activity into a person’s routine usually helps them maintain it. However, he’s identified two common problems:

• People lack understanding about their schedules and routines.
• People have schedules that fluctuate from one day to the next.

“Individuals face a lot of changes in their life,” Xu said. “Maybe they’re a student who has graduated, and they’re going into industry, which means their daily routine will be different from what it was while they were in school. This app allows them to experiment with different time slots and activity types to figure out another way and help them update their activity routine no matter what life changes they face.”

CUSTOM FIT

Some users who have been inactive for extended periods may be unsure how much exercise they need. Plannergy can also help them determine the intensity level of the activity to help avoid overexertion. 

“If someone has been inactive for months, it’s hard to ask them to run two miles daily,” Xu said. “There’s much for them to figure out. How much do they want to do, and at what intensity level? This app lets them gradually figure out the ideal activity. They can continue to track their progress and see if improvements are needed.”

Plannergy is not limited to physical activity. Xu says one of the students in his study who worked out daily used the app to identify times in her schedule to take breaks or focus on more spiritual disciplines.

“She added yoga and removed some high-intensity physical activities, and her sleeping routine also changed,” Xu said.

Xu is working to improve the app. Future versions will have sensing technology to leverage health informatics so users can make better decisions. He also wants the app to record user data and make customized suggestions for activities that fit the user’s schedule and preferred exercise intensity level.

“The app requires manual tracking, which can create user burden,” he said. “I think in the future, the process could be more automated. We want to keep it flexible but add more scaffolding to enhance user experience.”

Their study of language detection software found that algorithms struggle to differentiate anti-Asian violence-provoking speech from general hate speech. Left unchecked, threats of violence online can go unnoticed and turn into real-world attacks. 

Researchers from Georgia Tech and the Anti-Defamation League (ADL) teamed together in the study. They made their discovery while testing natural language processing (NLP) models trained on data they crowdsourced from Asian communities. 

“The Covid-19 pandemic brought attention to how dangerous violence-provoking speech can be. There was a clear increase in reports of anti-Asian violence and hate crimes,” said Gaurav Verma, a Georgia Tech Ph.D. candidate who led the study. 

“Such speech is often amplified on social platforms, which in turn fuels anti-Asian sentiments and attacks.”

Violence-provoking speech differs from more commonly studied forms of harmful speech, like hate speech. While hate speech denigrates or insults a group, violence-provoking speech implicitly or explicitly encourages violence against targeted communities.

Humans can define and characterize violent speech as a subset of hateful speech. However, computer models struggle to tell the difference due to subtle cues and implications in language.

The researchers tested five different NLP classifiers and analyzed their F1 score, which measures a model's performance. The classifiers reported a 0.89 score for detecting hate speech, while detecting violence-provoking speech was only 0.69. This contrast highlights the notable gap between these tools and their accuracy and reliability. 

The study stresses the importance of developing more refined methods for detecting violence-provoking speech. Internet misinformation and inflammatory rhetoric escalate tensions that lead to real-world violence. 

The Covid-19 pandemic exemplified how public health crises intensify this behavior, helping inspire the study. The group cited that anti-Asian crime across the U.S. increased by 339% in 2021 due to malicious content blaming Asians for the virus. 

The researchers believe their findings show the effectiveness of community-centric approaches to problems dealing with harmful speech. These approaches would enable informed decision-making between policymakers, targeted communities, and developers of online platforms.

Along with stronger models for detecting violence-provoking speech, the group discusses a direct solution: a tiered penalty system on online platforms. Tiered systems align penalties with severity of offenses, acting as both deterrent and intervention to different levels of harmful speech. 

“We believe that we cannot tackle a problem that affects a community without involving people who are directly impacted,” said Jiawei Zhou, a Ph.D. student who studies human-centered computing at Georgia Tech. 

“By collaborating with experts and community members, we ensure our research builds on front-line efforts to combat violence-provoking speech while remaining rooted in real experiences and needs of the targeted community.”

The researchers trained their tested NLP classifiers on a dataset crowdsourced from a survey of 120 participants who self-identified as Asian community members. In the survey, the participants labeled 1,000 posts from X (formerly Twitter) as containing either violence-provoking speech, hateful speech, or neither.

Since characterizing violence-provoking speech is not universal, the researchers created a specialized codebook for survey participants. The participants studied the codebook before their survey and used an abridged version while labeling. 

To create the codebook, the group used an initial set of anti-Asian keywords to scan posts on X from January 2020 to February 2023. This tactic yielded 420,000 posts containing harmful, anti-Asian language. 

The researchers then filtered the batch through new keywords and phrases. This refined the sample to 4,000 posts that potentially contained violence-provoking content. Keywords and phrases were added to the codebook while the filtered posts were used in the labeling survey.

The team used discussion and pilot testing to validate its codebook. During trial testing, pilots labeled 100 Twitter posts to ensure the sound design of the Asian community survey. The group also sent the codebook to the ADL for review and incorporated the organization’s feedback. 

“One of the major challenges in studying violence-provoking content online is effective data collection and funneling down because most platforms actively moderate and remove overtly hateful and violent material,” said Tech alumnus Rynaa Grover (M.S. CS 2024).

“To address the complexities of this data, we developed an innovative pipeline that deals with the scale of this data in a community-aware manner.”

Emphasis on community input extended into collaboration within Georgia Tech’s College of Computing. Faculty members Srijan Kumar and Munmun De Choudhury oversaw the research that their students spearheaded.

Kumar, an assistant professor in the School of Computational Science and Engineering, advises Verma and Grover. His expertise is in artificial intelligence, data mining, and online safety.

De Choudhury is an associate professor in the School of Interactive Computing and advises Zhou. Their research connects societal mental health and social media interactions.

The Georgia Tech researchers partnered with the ADL, a leading non-governmental organization that combats real-world hate and extremism. ADL researchers Binny Mathew and Jordan Kraemer co-authored the paper.

The group will present its paper at the 62nd Annual Meeting of the Association for Computational Linguistics (ACL 2024), which takes place in Bangkok, Thailand, Aug. 11-16 

ACL 2024 accepted 40 papers written by Georgia Tech researchers. Of the 12 Georgia Tech faculty who authored papers accepted at the conference, nine are from the College of Computing, including Kumar and De Choudhury.

“It is great to see that the peers and research community recognize the importance of community-centric work that provides grounded insights about the capabilities of leading language models,” Verma said. 

“We hope the platform encourages more work that presents community-centered perspectives on important societal problems.” 

Visit https://sites.gatech.edu/research/acl-2024/ for news and coverage of Georgia Tech research presented at ACL 2024.

They’re launching a new study focused on disability in the STEM fields of work — science, technology, engineering, and mathematics, which they hypothesize are a good fit for people with physical disabilities. Technology has made the work more accessible. Plus, the pay is good. However, there are challenges for working people with disabilities that even a great salary can’t overcome.

“We envision a scenario in which people with disabilities can get into the workforce and provide for their needs,” said Mitchell, associate professor in the Wallace H. Coulter Department of Biomedical Engineering.

Quinn, one of Mitchell’s former students, graduated in May with his master’s in computer science. He was well-known on campus for the sign attached to the back of his wheelchair, which said “THWG” — or “To Hell With Georgia” — a nod to the famous Georgia Tech-University of Georgia rivalry Quinn shares with his older sister, who attended UGA.

“The overall objective with this data-enabled study is to highlight the factors in academia and industry that have historically inhibited the successful inclusion of disabled people in STEM work,” said Quinn, who took the lead role in this study, which will gather data from both non-disabled and disabled people.

“We want to get a more complete picture of the current landscape, of the educational environment and the workplace,” said Mitchell, principal investigator of the Laboratory for Pathology Dynamics. 

Increasing the Sample Size

The study is part of the Science Leadership award Mitchell’s lab received in October 2022. This program, supported by the Chan Zuckerberg Initiative and the National Academies of Sciences, Engineering, and Medicine, supports early-career biomedical researchers who have a record of promoting diversity, equity, and inclusion. The award includes a $1.15 million grant over five years.

Mitchell, an internationally recognized Paralympian, developed a neurological condition as a teen that resulted in quadriplegia. She’s always made it a point in her lab to include students from diverse backgrounds and disabilities. 

“There is almost no data out there about the inclusion of disabled people in the workforce, only tiny sample sizes,” Mitchell said. “So we wanted to go after a larger sample size. Because if we are not reaching appropriate inclusion — and the few existing studies show that we’re not — then we want to know why.” 

Quinn added, “Stable and high-paying careers in STEM fields seem like a viable option for people with disabilities to both achieve and maintain financial independence.”

Grappling With the Disability Tax

For a person with significant disability, even a good-paying job may not be enough to offset the “disability tax.” Quinn defines the tax as “the extra time and money that living with a disability takes.”

For example, some people need a monthly disability check to cover common living expenses. But often, a more valuable government benefit is a health plan that covers “the thousands of dollars per month in personal support and care services,” Quinn explained. “You often only qualify for this if you’re on government disability benefits and making less than a certain amount of money per month.”

Also, policies vary by state, so individuals can easily fall through the cracks due to the complexities of various programs. And private or employer-funded healthcare plans typically can’t compete with government plans, which cover these expensive personal support services. 

For many people with disabilities, it comes down to a choice between working or government-supported services.

“There doesn’t seem to be a middle ground,” said Mitchell, who estimates approximately 60% of her income supports her medical and disability needs. “And that’s after insurance.”

The researchers hope their study provides momentum that will result in something close to full accessibility.

“This study will illuminate the challenges, even if it doesn’t solve them,” said Mitchell. “And while we’re focusing on STEM, this kind of study can be extrapolated to other fields as well. Whether you’re in science or not, I think people understand we’re asking important societal questions.”

Take the Survey

The Georgia Tech group combined two current ML methods into their hybrid model called MRM-GP (Multi-Region Markovian Gaussian Process). 

Neuroscientists who use MRM-GP learn more about communications and interactions within the brain. This in turn improves understanding of brain functions and disorders.

“Clinically, MRM-GP could enhance diagnostic tools and treatment monitoring by identifying and analyzing neural activity patterns linked to various brain disorders,” said Weihan Li, the study’s lead researcher. 

“Neuroscientists can leverage MRM-GP for its robust modeling capabilities and efficiency in handling large-scale brain data.” 

MRM-GP reveals where and how communication travels across brain regions. 

The group tested MRM-GP using spike trains and local field potential recordings, two kinds of measurements of brain activity. These tests produced representations that illustrated directional flow of communication among brain regions. 

Experiments also disentangled brainwaves, called oscillatory interactions, into organized frequency bands. MRM-GP’s hybrid configuration allows it to model frequencies and phase delays within the latent space of neural recordings.

MRM-GP combines the strengths of two existing methods: the Gaussian process (GP) and linear dynamical systems (LDS). The researchers say that MRM-GP is essentially an LDS that mirrors a GP.

LDS is a computationally efficient and cost-effective method, but it lacks the power to produce representations of the brain. GP-based approaches boost LDS's power, facilitating the discovery of variables in frequency bands and communication directions in the brain.

Converting GP outputs into an LDS is a difficult task in ML. The group overcame this challenge by instilling separability in the model’s multi-region kernel. Separability establishes a connection between the kernel and LDS while modeling communication between brain regions.

Through this approach, MRM-GP overcomes two challenges facing both neuroscience and ML fields. The model helps solve the mystery of intraregional brain communication. It does so by bridging a gap between GP and LDS, a feat not previously accomplished in ML.

“The introduction of MRM-GP provides a useful tool to model and understand complex brain region communications,” said Li, a Ph.D. student in the School of Computational Science and Engineering (CSE). 

“This marks a significant advancement in both neuroscience and machine learning.”

Fellow doctoral students Chengrui Li and Yule Wang co-authored the paper with Li. School of CSE Assistant Professor Anqi Wu advises the group. 

Each MRM-GP student pursues a different Ph.D. degree offered by the School of CSE. W. Li studies computer science, C. Li studies computational science and engineering, and Wang studies machine learning. The school also offers Ph.D. degrees in bioinformatics and bioengineering.

Wu is a 2023 recipient of the Sloan Research Fellowship for neuroscience research. Her work straddles two of the School’s five research areas: machine learning and computational bioscience. 

MRM-GP will be featured at the world’s top conference on ML and artificial intelligence. The group will share their work at the International Conference on Machine Learning (ICML 2024), which will be held July 21-27 in Vienna. 

ICML 2024 also accepted for presentation a second paper from Wu’s group intersecting neuroscience and ML. The same authors will present A Differentiable Partially Observable Generalized Linear Model with Forward-Backward Message Passing.

Twenty-four Georgia Tech faculty from the Colleges of Computing and Engineering will present 40 papers at ICML 2024. Wu is one of six faculty representing the School of CSE who will present eight total papers.

The group’s ICML 2024 presentations exemplify Georgia Tech’s focus on neuroscience research as a strategic initiative.  

Wu is an affiliated faculty member with the Neuro Next Initiative, a new interdisciplinary program at Georgia Tech that will lead research in neuroscience, neurotechnology, and society. The University System of Georgia Board of Regents recently approved a new neuroscience and neurotechnology Ph.D. program at Georgia Tech. 

“Presenting papers at international conferences like ICML is crucial for our group to gain recognition and visibility, facilitates networking with other researchers and industry professionals, and offers valuable feedback for improving our work,” Wu said. 

“It allows us to share our findings, stay updated on the latest developments in the field, and enhance our professional development and public speaking skills.”

Visit https://sites.gatech.edu/research/icml-2024 for news and coverage of Georgia Tech research presented at ICML 2024.

The Tech group dubbed its method PRODIGY (PROjected DIffusion for controlled Graph Generation). PRODIGY enables diffusion models to generate 3D images of complex structures, such as molecules from chemical formulas. 

Scientists in pharmacology, materials science, social network analysis, and other fields can use PRODIGY to simulate large-scale networks. By generating 3D molecules from multiple graph datasets, the group proved that PRODIGY could handle complex structures.

In keeping with its name, PRODIGY is the first plug-and-play machine learning (ML) approach to controllable graph generation in diffusion models. This method overcomes a known limitation inhibiting diffusion models from broad use in science and engineering.

“We hope PRODIGY enables drug designers and scientists to generate structures that meet their precise needs,” said Kartik Sharma, lead researcher on the project. “It should also inspire future innovations to precisely control modern generative models across domains.” 

PRODIGY works on diffusion models, a generative AI model for computer vision tasks. While suitable for image creation and denoising, diffusion methods are limited because they cannot accurately generate graph representations of custom parameters a user provides.

PRODIGY empowers any pre-trained diffusion model for graph generation to produce graphs that meet specific, user-given constraints. This capability means, as an example, that a drug designer could use any diffusion model to design a molecule with a specific number of atoms and bonds.

The group tested PRODIGY on two molecular and five generic datasets to generate custom 2D and 3D structures. This approach ensured the method could create such complex structures, accounting for the atoms, bonds, structures, and other properties at play in molecules. 

Molecular generation experiments with PRODIGY directly impact chemistry, biology, pharmacology, materials science, and other fields. The researchers say PRODIGY has potential in other fields using large networks and datasets, such as social sciences and telecommunications.

These features led to PRODIGY’s acceptance for presentation at the upcoming International Conference on Machine Learning (ICML 2024). ICML 2024 is the leading international academic conference on ML. The conference is taking place July 21-27 in Vienna.

Assistant Professor Srijan Kumar is Sharma’s advisor and paper co-author. They worked with Tech alumnus Rakshit Trivedi (Ph.D. CS 2020), a Massachusetts Institute of Technology postdoctoral associate.

Twenty-four Georgia Tech faculty from the Colleges of Computing and Engineering will present 40 papers at ICML 2024. Kumar is one of six faculty representing the School of Computational Science and Engineering (CSE) at the conference.

Sharma is a fourth-year Ph.D. student studying computer science. He researches ML models for structured data that are reliable and easily controlled by users. While preparing for ICML, Sharma has been interning this summer at Microsoft Research in the Research for Industry lab.

“ICML is the pioneering conference for machine learning,” said Kumar. “A strong presence at ICML from Georgia Tech illustrates the ground-breaking research conducted by our students and faculty, including those in my research group.”

Visit https://sites.gatech.edu/research/icml-2024 for news and coverage of Georgia Tech research presented at ICML 2024.

Since humans often associate objects in a home with the room they are in, Naoki Yokoyama thinks robots that navigate human environments to perform assistive tasks should mimic that reasoning.

Roboticists have employed natural language models to help robots mimic human reasoning over the past few years. However, Yokoyama, a Ph.D. student in robotics, said these models create a “bottleneck” that prevents agents from picking up on visual cues such as room type, size, décor, and lighting. 

Yokoyama presented a new framework for semantic reasoning at the Institute of Electrical and Electronic Engineers (IEEE) International Conference on Robotics and Automation (ICRA) last month in Yokohama, Japan. ICRA is the world’s largest robotics conference.

Yokoyama earned a best paper award in the Cognitive Robotics category with his Vision-Language Frontier Maps (VLFM) proposal.

Assistant Professor Sehoon Ha and Associate Professor Dhruv Batra from the School of Interactive Computing advised Yokoyama on the paper. Yokoyama authored the paper while interning at the Boston Dynamics’ AI Institute.

“I think the cognitive robotic category represents a significant portion of submissions to ICRA nowadays,” said Yokoyama, whose family is from Japan. “I’m grateful that our work is being recognized among the best in this field.”

Instead of natural language models, Yokoyama used a renowned vision-language model called BLIP-2 and tested it on a Boston Dynamics “Spot” robot in home and office environments.

“We rely on models that have been trained on vast amounts of data collected from the web,” Yokoyama said. “That allows us to use models with common sense reasoning and world knowledge. It’s not limited to a typical robot learning environment.”

What is Blip-2?

BLIP-2 matches images to text by assigning a score that evaluates how well the user input text describes the content of an image. The model removes the need for the robot to use object detectors and language models. 

Instead, the robot uses BLIP-2 to extract semantic values from RGB images with a text prompt that includes the target object. 

BLIP-2 then teaches the robot to recognize the room type, distinguishing the living room from the bathroom and the kitchen. The robot learns to associate certain objects with specific rooms where it will likely find them.

From here, the robot creates a value map to determine the most likely locations for a target object, Yokoyama said.

Yokoyama said this is a step forward for intelligent home assistive robots, enabling users to find objects — like missing keys — in their homes without knowing an item’s location. 

“If you’re looking for a pair of scissors, the robot can automatically figure out it should head to the kitchen or the office,” he said. “Even if the scissors are in an unusual place, it uses semantic reasoning to work through each room from most probable location to least likely.”

He added that the benefit of using a VLM instead of an object detector is that the robot will include visual cues in its reasoning.

“You can look at a room in an apartment, and there are so many things an object detector wouldn’t tell you about that room that would be informative,” he said. “You don’t want to limit yourself to a textual description or a list of object classes because you’re missing many semantic visual cues.”

While other VLMs exist, Yokoyama chose BLIP-2 because the model:

• Accepts any text length and isn’t limited to a small set of objects or categories.
• Allows the robot to be pre-trained on vast amounts of data collected from the internet.
• Has proven results that enable accurate image-to-text matching.

Home, Office, and Beyond

Yokoyama also tested the Spot robot to navigate a more challenging office environment. Office spaces tend to be more homogenous and harder to distinguish from one another than rooms in a home. 

“We showed a few cases in which the robot will still work,” Yokoyama said. “We tell it to find a microwave, and it searches for the kitchen. We tell it to find a potted plant, and it moves toward an area with windows because, based on what it knows from BLIP-2, that’s the most likely place to find the plant.”

Yokoyama said as VLM models continue to improve, so will robot navigation. The increase in the number of VLM models has caused robot navigation to steer away from traditional physical simulations.

“It shows how important it is to keep an eye on the work being done in computer vision and natural language processing for getting robots to perform tasks more efficiently,” he said. “The current research direction in robot learning is moving toward more intelligent and higher-level reasoning. These foundation models are going to play a key role in that.”

Top photo by Kevin Beasley/College of Computing.