The National Science Foundation (NSF) awarded four researchers at the George Washington University its Faculty Early Career Development Program (CAREER) award, the federal agency’s most prestigious honor supporting new work from junior faculty with the potential to serve as academic role models in research and education and to lead advances in their fields, in the 2023 fiscal year. The projects are funded over five years.
Currently, 15 GW faculty have active CAREER awards, including the four new recipients. Among the 15 active awardees are three GW faculty who received their CAREER awards at previous institutions before moving in FY23 to GW, where they are continuing their CAREER-funded research. These include GW Engineering Professors Caitlin Grady and Mathew Rau, both of whom received their CAREER awards at Penn State, and GW Computer Science Professor Sibin Mohan, who received his CAREER award at the University of Illinois before moving to GW.
"A diverse and growing number of awardees at GW shows that we are not just fostering top faculty talent but are attracting world-class investigators to join our research community," Pam Norris, vice provost for research at GW and a former CAREER awardee, said. "My own award prepared me for leadership both in the lab and in the classroom. These faculty have already demonstrated early success in their careers and are well on their way to becoming among the most impactful scholars in their fields."
The four new CAREER awardees represent a range of disciplines, from computer science and chemistry to statistics and engineering, and their work focuses on everything from the very small (mitochondria producing energy for brain cells) to the unfathomably large (corporate databases of private data). Learn about the newest awardees and their projects below:
Gina Adam
Building reliable, brain-inspired computer chips
Innovators in the field of artificial intelligence face a consistent challenge: the enormous amount of computing power required to train AIs, as well as the significant hardware, environmental and financial resources these complex programs consume. Compact, efficient memristor chips and other emerging technologies have demonstrated promise to support more energy-efficient AI algorithms. But these technologies are currently early in development and suffer from variability. No two devices are alike, which makes it difficult to incorporate them in commercial products.
This is a grand challenge that Gina Adam, an assistant professor in the Electrical and Computer Engineering Department at GW's School of Engineering and Applied Science (GW Engineering) is currently tackling thanks to her NSF CAREER award. Adam and her team are developing the next generation of ultra-low variability memristors, providing the academic performance boost these technologies need for future commercial adoption. Because Adam’s students help develop and test the memristor prototypes, and because of the lab’s outreach to local high school students, the project also helps develop the next generation of a key workforce—microelectronics designers, researchers and engineers.
Ling Hao
Illuminating how mitochondria work in the brain
Neurons in the brain depend on mitochondria, the famous “powerhouses of the cell,” to meet the enormous energy demands of cognitive activity. But major technological limitations stand in the way of understanding the specific molecular microenvironment of these neuronal mitochondria. Ling Hao, an assistant professor in the Columbian College of Arts and Sciences (CCAS) Department of Chemistry, is pushing those technological limits by developing mass spectrometry techniques that will help researchers capture and analyze neuronal mitochondrial activities and interactions at the molecular level.
In partnership with Thermo Fisher and Shimazu scientific instrument companies and Washington-Baltimore Mass Spectrometry Discussion group, Hao and her team will also organize a series of education and outreach events focused on mass spectrometry technology and applications for local students and scientists.
Fang Jin
Eliminating AI bias in medical imaging
Trained deep learning models—AIs that perform predictive tasks in language processing, image analysis and other fields—have millions or billions of parameters and can produce highly accurate results, sometimes surpassing human expert performance. However, the opacity of these models’ decision-making processes raises concerns about their fairness and trustworthiness. Researchers evaluating the validity of an AI-classified tumor in a target medical image need to understand the images on which that model trained, for instance.
Fang Jin, an assistant professor in CCAS’ Department of Statistics, will lead a project designing and developing a universal interpretation framework that will enable humans to understand the decision-making process of increasingly complex black-box Deep Neural Networks (DNNs) trained on medical images, videos, natural language processing and deep reinforcement learning. The interpretation framework will produce feedback on what scientific knowledge these DNNs currently perceive, helping researchers refine their models by identifying and minimizing bias within them. Jin and her team also will lead community outreach initiatives, including a professional development program for K-12 teachers and a deep learning summer camp for high school students.
Arkady Yerukhimovich
Creating more robust data protection protocols
Attention, internet shoppers: Sensitive personal data is everywhere online, and a growing number of popular apps and services depend on collecting and analyzing large amounts of it. In the face of the unprecedented privacy concerns raised by this computing paradigm, a key research challenge is to design protocols for secure and private computation that can scale to these massive volumes of data. Arkady Yerukhimovich, an assistant professor of computer engineering at GW Engineering, is developing such protocols.
Yerukhimovich and his team combine techniques from secure multi-party computation, sketching algorithms and differential privacy—a database security concept that limits the personal information revealed through aggregate computations—to create protocols that protect individual users and are robust against malicious actors without running up massive computation and communication costs. The project will enable new secure computations for large-data applications like machine learning and network measurement, while also producing research opportunities and course materials for graduate and undergraduate students.
