Dutta Receives NSF CAREER Award to Advance Reliable and Trustworthy AI
Sanghamitra Dutta, an assistant professor of electrical and computer engineering, has received funding from the National Science Foundation (NSF) to continue her work in advancing reliable and trustworthy artificial intelligence (AI) for social good.
Dutta—who has affiliate appointments in the University of Maryland Institute for Advanced Computer Studies and the University of Maryland Center for Machine Learning—is principal investigator of an NSF Faculty Early Career Development Program (CAREER) award, expected to total $665,550 over the next five years.
This highly competitive award, considered one of NSF’s most prestigious awards in support of early-career faculty, is given to researchers who have the potential to serve as academic role models in research and education and to lead advances in the mission of their department or organization.
Dutta’s project seeks to advance the foundations of ethical and socially responsible machine learning by empowering users to systematically identify, explain and mitigate various sources of disparity. Rethinking the traditional paradigm of separately addressing fairness and explainability, her work will jointly examine fairness and explainability through a unified information-theoretic lens.
A key element of the project, Dutta says, is the development of a novel information-theoretic view of responsible machine learning, leveraging a body of work in information theory called Partial Information Decomposition (PID).
Four distinct research thrusts will be investigated, Dutta says. The first is providing an information-theoretic framework for explaining sources of disparity with respect to protected attributes such as gender and race. Second is performing a systematic feature selection and representation learning with disparity control. Third is investigating fundamental limits with a focus on distributed and federated settings. And fourth is validating the findings on real-world datasets in finance and education.
Additionally, through extensive outreach and student engagements, the NSF-funded project aims to instill interest in mathematically principled approaches and STEM education among students to spearhead the next generation of socially responsible technology development.
A UMD faculty member since 2022, Dutta has published actively in several leading machine learning conferences. A recent paper from Dutta’s research group on explainability, “Robust Counterfactual Explanations for Neural Networks with Probabilistic Guarantees," was presented at the 2023 International Conference on Machine Learning (ICML).
Another recent work from her group, “Demystifying Local and Global Fairness Trade-offs in Federated Learning Using Partial Information Decomposition,” will be presented at the upcoming 2024 International Conference on Learning Representations (ICLR). This work was also featured in the Montreal AI Ethics Brief.
***
“Information-Theoretic Measures for Fairness and Explainability in High-Stakes Applications” is supported by NSF grant #2340006 from the NSF’s Division of Computing and Communication Foundations.
PI: Sanghamitra Dutta, an assistant professor of electrical and computer engineering with affiliate appointments in the University of Maryland Institute for Advanced Computer Studies and the University of Maryland Center for Machine Learning.
About the CAREER award: The Faculty Early Career Development (CAREER) Program is an NSF activity that offers the foundation’s most prestigious awards in support of junior faculty who exemplify the role of teacher-scholars through outstanding research, excellent education and the integration of education and research within the context of the mission of their organization.