USC + Amazon Center on Secure and Trusted Machine Learning selects six new research projects
Faculty and academic-fellow projects are focused on various aspects of trustworthy machine learning.
The USC + Amazon Center on Secure and Trusted Machine Learning, established in January 2021 to support fundamental research and development of new approaches to machine learning (ML) privacy, security, and trustworthiness, today announced it has selected six research projects for 2023-2024, including four faculty projects and two academic projects led by new Amazon ML Fellows.
“The USC-Amazon partnership continues to flourish,” said Yannis C. Yortsos, dean of the USC Viterbi School of Engineering. “The selection of the new projects and fellows is another testament to the quality and strength of our partnership. We look forward to their successful and impactful implementation for the collective, societal benefit.”
Below is information about the six research project recipients, including their areas of research:
- Salman Avestimehr, Dean’s Professor of electrical and computer engineering and computer science: “Advancing continual and federated learning with self- and mixed supervision.” Avestimehr’s research goal is to better leverage unlabeled data under the data heterogeneity and privacy constraints of distributed-learning settings, using two different approaches.
- Muhao Chen, research assistant professor of computer science: “Robust (controlled) natural-language generation with structure‐aware equivariance learning.” Chen’s research seeks to address challenges inherent in using pretrained sequence-to-sequence models for controlled natural-language generation (NLG), using a novel NLG framework based on structural-equivariance learning.
- Lars Lindemann, assistant professor of computer science, and Soutirini Chattpadhyay, assistant professor of computer science: “Still don’t trust me? Building trustworthy AI code generation.” Their research proposes a new framework that decomposes intended user behavior into a sequence of code suggestions and adapts dynamically to maintain trust.
- Konstantinos Psounis, professor of electrical and computer engineering and computer science: “Private labeling and learning for voice assistants with cameras.” Psounis’s research will take on the challenge of achieving and maintaining the privacy of user videos using a variety of approaches.
- Brihi Joshi is a third-year computer science PhD student, advised by Xiang Ren. Joshi’s research goal is to incorporate human-centered explainability into NLP systems, inspired by how humans interact among themselves and with these systems, while also borrowing insights from other disciplines.
- Fei Wang is a PhD candidate in the computer science department, advised by Chen. His research interests lie in natural-language processing and machine learning, and his long-term goal is to build robust, controllable, and accountable large-language-model-based systems.
“Generative AI has recently ignited a profound set of questions to understand the societal impacts of AI,” said Avestimehr, the inaugural director of the USC + Amazon Center, who is also an Amazon Scholar. “Investigations into issues such as privacy, security, trustworthiness, and ownership have gained unprecedented attention, underscoring our vision when we started the center three years ago.”