In the October 2025 PG semester, I am teaching the course EEE 6617 Privacy Preserving Machine Learning. The course outline is below:

Review of common machine learning algorithms; mathematical definition of privacy; case studies of high-profile privacy breaches; common attack and threat models; differential privacy (DP); other privacy approaches; basic building blocks of privacy-preserving algorithm design; achieving DP via noise for numeric queries; achieving DP via sampling for non-numeric queries; the Gaussian mechanism; composition of multi-stage differentially private algorithms; differentially private empirical risk minimization; differentially private stochastic gradient descent; differential privacy for distributed data and federated learning; differential privacy for neural networks; local differential privacy; differentially private heavy hitters; emerging applications; machine unlearning.