Postdoc, University of Wisconsin - Madison
1 paper at NeurIPS 2025
We develop efficient, differentially private algorithms for bilevel optimization with near-optimal excess risk bounds that nearly match the best rates for standard private optimization.