A Georgia Tech alum’s dissertation introduced ways to make artificial intelligence (AI) more accessible, interpretable, and accountable. Although it’s been a year since his doctoral defense, Zijie (Jay) Wang’s (Ph.D. ML-CSE 2024) work continues to resonate with researchers.

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.”

Many communities rely on insights from computer-based models and simulations. This week, a nest of Georgia Tech experts are swarming an international conference to present their latest advancements in these tools, which offer solutions to pressing challenges in science and engineering.

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]

Many communities rely on insights from computer-based models and simulations. This week, a nest of Georgia Tech experts are swarming an international conference to present their latest advancements in these tools, which offer solutions to pressing challenges in science and engineering.

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]

Men and women in California put their lives on the line when battling wildfires every year, but there is a future where machines powered by artificial intelligence are on the front lines, not firefighters.

However, this new generation of self-thinking robots would need security protocols to ensure they aren’t susceptible to hackers. To integrate such robots into society, they must come with assurances that they will behave safely around humans.

It begs the question: can you guarantee the safety of something that doesn’t exist yet? It’s something Assistant Professor Glen Chou hopes to accomplish by developing algorithms that will enable autonomous systems to learn and adapt while acting with safety and security assurances. 

He plans to launch research initiatives, in collaboration with the School of Cybersecurity and Privacy and the Daniel Guggenheim School of Aerospace Engineering, to secure this new technological frontier as it develops. 

“To operate in uncertain real-world environments, robots and other autonomous systems need to leverage and adapt a complex network of perception and control algorithms to turn sensor data into actions,” he said. “To obtain realistic assurances, we must do a joint safety and security analysis on these sensors and algorithms simultaneously, rather than one at a time.”

This end-to-end method would proactively look for flaws in the robot’s systems rather than wait for them to be exploited. This would lead to intrinsically robust robotic systems that can recover from failures.

Chou said this research will be useful in other domains, including advanced space exploration. If a space rover is sent to one of Saturn’s moons, for example, it needs to be able to act and think independently of scientists on Earth. 

Aside from fighting fires and exploring space, this technology could perform maintenance in nuclear reactors, automatically maintain the power grid, and make autonomous surgery safer. It could also bring assistive robots into the home, enabling higher standards of care. 

This is a challenging domain where safety, security, and privacy concerns are paramount due to frequent, close contact with humans.

This will start in the newly established Trustworthy Robotics Lab at Georgia Tech, which Chou directs. He and his Ph.D. students will design principled algorithms that enable general-purpose robots and autonomous systems to operate capably, safely, and securely with humans while remaining resilient to real-world failures and uncertainty.

Chou earned dual bachelor’s degrees in electrical engineering and computer sciences as well as mechanical engineering from University of California Berkeley in 2017, a master’s and Ph.D. in electrical and computer engineering from the University of Michigan in 2019 and 2022, respectively. He was a postdoc at MIT Computer Science & Artificial Intelligence Laboratory prior to joining Georgia Tech in November 2024. He is a recipient of the National Defense Science and Engineering Graduate fellowship program, NSF Graduate Research fellowships, and was named a Robotics: Science and Systems Pioneer in 2022.

Calculating and visualizing a realistic trajectory of ink spreading through water has been a longstanding and enormous challenge for computer graphics and physics researchers.

When a drop of ink hits the water, it typically sinks forward, creating a tail before various ink streams branch off in different directions. The motion of the ink’s molecules upon mixing with water is seemingly random. This is because the motion is determined by the interaction of the water’s viscosity (thickness) and vorticity (how much it rotates at a given point).

“If the water is more viscous, there will be fewer branches. If the water is less viscous, it will have more branches,” said Zhiqi Li, a graduate computer science student.

Li is the lead author of Particle-Laden Fluid on Flow Maps, a best paper winner at the December 2024 ACM SIGGRAPH Asia conference. Assistant Professor Bo Zhu advises Li and is the co-author of six papers accepted to the conference.

Zhu said they must correctly calculate and simulate the interaction between viscosity and vorticity before they can accurately predict the ink trajectory.

“The ink branches generate based on the intricate interaction between the vorticities and the viscosity over time, which we simulated,” Zhu said. “Using a standard method to simulate the physics will cause most of the structures to fade quickly without being able to see any detailed hierarchies.”

Zhu added that researchers had yet to develop a method for this until he and his co-authors proposed a new way to solve the equation. Their breakthrough has unlocked the most accurate simulations of ink diffusion to date.

“Ink diffusion is one of the most visually striking examples of particle-laden flow,” Zhu said.

“We introduce a new viscosity model that solves for the interaction between vorticity and viscosity from a particle flow map perspective. This new simulation lets you map physical quantities from a certain time frame, allowing us to see particle trajectory.”

In computer simulations, flow is the digital visualization of a gas or liquid through a system. Users can simulate these liquids and gases through different scenarios and study pressure, velocity, and temperature.

A particle-laden flow depicts solid particles mixing within a continuous fluid phase, such as dust or water sediment. A flow map traces particle motion from the start point to the endpoint.

Duowen Chen, a computer science Ph.D. student also advised by Zhu and co-author of the paper, said previous efforts by researchers to simulate ink diffusion depended on guesswork. They either used limited traditional methods of calculations or artificial designs. 

“They add in a noise model or an artificial model to create vortical motions, but our method does not require adding any artificial vortical components,” Chen said. “We have a better viscosity force calculation and vortical preservation, and the two give a better ink simulation.”

Zhu also won a best paper award at the 2023 SIGGRAPH Asia conference for his work explaining how neural network maps created through artificial intelligence (AI) could close the gaps of difficult-to-solve equations. In his new paper, he said it was essential to find a way to simulate ink diffusion accurately independent of AI.

“If we don’t have to train a large-scale neural network, then the computation time will be much faster, and we can reduce the computation and memory costs,” Zhu said. “The particle flow map representation can preserve those particle structures better than the neural network version, and they are a widely used data structure in traditional physics-based simulation.”

Georgia Tech researchers have created a dataset that trains computer models to understand nuances in human speech during financial earnings calls. The dataset provides a new resource to study how public correspondence affects businesses and markets. 

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.”

A new machine learning (ML) model from Georgia Tech could protect communities from diseases, better manage electricity consumption in cities, and promote business growth, all at the same time.

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.”

Deven Desai and Mark Riedl have seen the signs for a while. 

Two years since OpenAI introduced ChatGPT, dozens of lawsuits have been filed alleging technology companies have infringed copyright by using published works to train artificial intelligence (AI) models.

Academic AI research efforts could be significantly hindered if courts rule in the plaintiffs' favor. 

Desai and Riedl are Georgia Tech researchers raising awareness about how these court rulings could force academic researchers to construct new AI models with limited training data. The two collaborated on a benchmark academic paper that examines the landscape of the ethical issues surrounding AI and copyright in industry and academic spaces.

“There are scenarios where courts may overreact to having a book corpus on your computer, and you didn’t pay for it,” Riedl said. “If you trained a model for an academic paper, as my students often do, that’s not a problem right now. The courts could deem training is not fair use. That would have huge implications for academia.

“We want academics to be free to do their research without fear of repercussions in the marketplace because they’re not competing in the marketplace,” Riedl said. 

Desai is the Sue and John Stanton Professor of Business Law and Ethics at the Scheller College of Business. He researches how business interests and new technology shape privacy, intellectual property, and competition law. Riedl is a professor at the College of Computing’s School of Interactive Computing, researching human-centered AI, generative AI, explainable AI, and gaming AI. 

Their paper, Between Copyright and Computer Science: The Law and Ethics of Generative AI, was published in the Northwestern Journal of Technology and Intellectual Property on Monday.

Desai and Riedl say they want to offer solutions that balance the interests of various stakeholders. But that requires compromise from all sides.

Researchers should accept they may have to pay for the data they use to train AI models. Content creators, on the other hand, should receive compensation, but they may need to accept less money to ensure data remains affordable for academic researchers to acquire.

Who Benefits?

The doctrine of fair use is at the center of every copyright debate. According to the U.S. Copyright Office, fair use permits the unlicensed use of copyright-protected works in certain circumstances, such as distributing information for the public good, including teaching and research.

Fair use is often challenged when one or more parties profit from published works without compensating the authors.

Any original published content, including a personal website on the internet, is protected by copyright. However, copyrighted material is republished on websites or posted on social media innumerable times every day without the consent of the original authors. 

In most cases, it’s unlikely copyright violators gained financially from their infringement.

But Desai said business-to-business cases are different. The New York Times is one of many daily newspapers and media companies that have sued OpenAI for using its content as training data. Microsoft is also a defendant in The New York Times’ suit because it invested billions of dollars into OpenAI’s development of AI tools like ChatGPT.

“You can take a copyrighted photo and put it in your Twitter post or whatever you want,” Desai said. “That’s probably annoying to the owner. Economically, they probably wanted to be paid. But that’s not business to business. What’s happening with Open AI and The New York Times is business to business. That’s big money.”

OpenAI started as a nonprofit dedicated to the safe development of artificial general intelligence (AGI) — AI that, in theory, can rival human thinking and possess autonomy.

These AI models would require massive amounts of data and expensive supercomputers to process that data. OpenAI could not raise enough money to afford such resources, so it created a for-profit arm controlled by its parent nonprofit.

Desai, Riedl, and many others argue that OpenAI ceased its research mission for the public good and began developing consumer products. 

“If you’re doing basic research that you’re not releasing to the world, it doesn’t matter if every so often it plagiarizes The New York Times,” Riedl said. “No one is economically benefitting from that. When they became a for-profit and produced a product, now they were making money from plagiarized text.”

OpenAI’s for-profit arm is valued at $80 billion, but content creators have not received a dime since the company has scraped massive amounts of copyrighted material as training data.

The New York Times has posted warnings on its sites that its content cannot be used to train AI models. Many other websites offer a robot.txt file that contains instructions for bots about which pages can and cannot be accessed. 

Neither of these measures are legally binding and are often ignored.

Solutions

Desai and Riedl offer a few options for companies to show good faith in rectifying the situation.

• Spend the money. Desai says Open AI and Microsoft could have afforded its training data and avoided the hassle of legal consequences.
  
  “If you do the math on the costs to buy the books and copy them, they could have paid for them,” he said. “It would’ve been a multi-million dollar investment, but they’re a multi-billion dollar company.”
   
• Be selective. Models can be trained on randomly selected texts from published works, allowing the model to understand the writing style without plagiarizing. 
  
  “I don’t need the entire text of War and Peace,” Desai said. “To capture the way authors express themselves, I might only need a hundred pages. I’ve also reduced the chance that my model will cough up entire texts.”
   
• Leverage libraries. The authors agree libraries could serve as an ideal middle ground as a place to store published works and compensate authors for access to those works, though the amount may be less than desired.
  
  “Most of the objections you could raise are taken care of,” Desai said. “They are legitimate access copies that are secure. You get access to only as much as you need. Libraries at universities have already become schools of information.”

Desai and Riedl hope the legal action taken by publications like The New York Times will send a message to companies that develop AI tools to pump the breaks. If they don’t, researchers uninterested in profit could pay the steepest price.

The authors say it’s not a new problem but is reaching a boiling point.

“In the history of copyright, there are ways that society has dealt with the problem of compensating creators and technology that copies or reduces your ability to extract money from your creation,” Desai said. “We wanted to point out there’s a way to get there.”

The Automatic Speech Recognition (ASR) models that power voice assistants like Amazon Alexa may have difficulty transcribing English speakers with minority dialects.

A study by Georgia Tech and Stanford researchers compared the transcribing performance of leading ASR models for people using Standard American English (SAE) and three minority dialects — African American Vernacular English (AAVE), Spanglish, and Chicano English.

Interactive Computing Ph.D. student Camille Harris is the lead author of a paper accepted into the 2024 Conference on Empirical Methods in Natural Language Processing (EMNLP) this week in Miami.

Harris recruited people who spoke each dialect and had them read from a Spotify podcast dataset, which includes podcast audio and metadata. Harris then used three ASR models — wav2vec 2.0, HUBERT, and Whisper — to transcribe the audio and compare their performances.

For each model, Harris found SAE transcription significantly outperformed each minority dialect. The models more accurately transcribed men who spoke SAE than women who spoke SAE. Members who spoke Spanglish and Chicano English had the least accurate transcriptions out of the test groups. 

While the models transcribed SAE-speaking women less accurately than their male counterparts, that did not hold true across minority dialects. Minority men had the most inaccurate transcriptions of all demographics in the study.

“I think people would expect if women generally perform worse and minority dialects perform worse, then the combination of the two must also perform worse,” Harris said. “That’s not what we observed. 

“Sometimes minority dialect women performed better than Standard American English. We found a consistent pattern that men of color, particularly Black and Latino men, could be at the highest risk for these performance errors.”

Addressing underrepresentation

Harris said the cause of that outcome starts with the training data used to build these models. Model performance reflected the underrepresentation of minority dialects in the data sets.

AAVE performed best under the Whisper model, which Harris said had the most inclusive training data of minority dialects.

Harris also looked at whether her findings mirrored existing systems of oppression. Black men have high incarceration rates and are one of the people groups most targeted by police. Harris said there could be a correlation between that and the low rate of Black men enrolled in universities, which leads to less representation in technology spaces.

“Minority men performing worse than minority women doesn’t necessarily mean minority men are more oppressed,” she said. “They may be less represented than minority women in computing and the professional sector that develops these AI systems.”

Harris also had to be cautious of a few variables among AAVE, including code-switching and various regional subdialects.

Harris noted in her study there were cases of code-switching to SAE. Speakers who code-switched performed better than speakers who did not. 

Harris also tried to include different regional speakers.

“It’s interesting from a linguistic and history perspective if you look at migration patterns of Black folks — perhaps people moving from a southern state to a northern state over time creates different linguistic variations,” she said. “There are also generational variations in that older Black Americans may speak differently from younger folks. I think the variation was well represented in our data. We wanted to be sure to include that for robustness.”

TikTok barriers

Harris said she built her study on a paper she authored that examined user-design barriers and biases faced by Black content creators on TikTok. She presented that paper at the Association of Computing Machinery’s (ACM) 2023 Conference on Computer Supported Cooperative Works. 

Those content creators depended on TikTok for a significant portion of their income. When providing captions for videos grew in popularity, those creators noticed the ASR tool built into the app inaccurately transcribed them. That forced the creators to manually input their captions, while SAE speakers could use the ASR feature to their benefit.

“Minority users of these technologies will have to be more aware and keep in mind that they’ll probably have to do a lot more customization because things won’t be tailored to them,” Harris said.

Harris said there are ways that designers of ASR tools could work toward being more inclusive of minority dialects, but cultural challenges could arise.

“It could be difficult to collect more minority speech data, and you have to consider consent with that,” she said. “Developers need to be more community-engaged to think about the implications of their models and whether it’s something the community would find helpful.”