Privacy-preserving probabilistic machine learning: a preview

Speaker: Nianqiao “Phyllis” Ju (Dartmouth)

Date: 10/28/25

Abstract: This talk is a preview of Math 146 in Winter 2026, which will focus on privacy-preserving probabilistic machine learning. The central goal is to learn about populations without revealing any sensitive information about any individual. We will give an accessible introduction to the definition of differential privacy, the randomized response mechanism, and the high-level ideas of differentially private optimization and sampling. No specialized background beyond probability and basic data analysis or machine learning is required. The goal is to provide a clear picture of what is currently possible and key open problems for further study. This will be a mostly nontechnical chalk talk.