Research Programs & Support
Current and previous research programs, grants, fellowships, and selected awards.
Current research programs
NSF RTG: Statistics and Data Theory — Engaging the Future of Data Science
Research and training in the mathematical and statistical foundations of data science and AI.
2026–2031 · Lead PI.
NSF AIMing: Synergistic Advancement of AI and Mathematics
Geometry of mathematical reasoning spaces; methods for AI systems that reason creatively about mathematics; and applications of AI to advance mathematical discovery.
2026–2028.
DARPA AIQ: Constraints for Provable Extrapolation in AI
Verification and certification of AI systems; function-level constraints, robustness verification, and extrapolation.
2025–2026.
NSF CAREER: Neural Networks in the Practical Regime
Moderate overparametrization, function space descriptions, data-dependent, explicit description of optimization biases.
2022–2027.
NSF MoDL: Occam's Razor in Deep and Physical Learning
Simplicity biases in neural network learning, including how optimization dynamics and model structure select low-complexity solutions.
Current through 2027.
DFG SPP 2298 Theoretical Foundations of Deep Learning: Combinatorial and Implicit Approaches to Deep Learning
Polyhedral geometry of neural networks; semi-algebraic descriptions of neural network function spaces; subdivisions of parameter spaces.
Current through 2027.
Additional and previous research support
BMFTR/DAAD SECAI: School of Embedded Composite Artificial Intelligence
A Zuse School advancing AI research and higher education through graduate training, research support, and academic–industrial exchange.
ERC Starting Grant: Deep Learning Theory
Theoretical foundations of deep learning, addressing capacity, optimization, and regularization through geometry, information geometry, and algebraic statistics.
2018–2023.
Selected fellowships & distinctions
UCLA Faculty Career Development Award
2021.