Assistant Professor, Rutgers University
2 papers at NeurIPS 2025
A method for continually and privately estimating both the first and second moments of a data stream with reduced noise compared to naive approaches.