This course will cover topics in two areas under privacy-preserving computing: a) federated or privacy-preserving
statistics and machine learning, and b) privacy-preserving outsourcing of computation to untrusted clouds. The techniques covered will include federated analytics and homomorphic encryption. The course will have a mix of theory and systems building. For example, it will discuss how federated analytics can be applied to learn signals across siloed hospital databases, and how homomorphic encryption is applied to real systems such as databases, media streaming services, anonymous messaging services, and machine learning systems.
There are no official prerequisites; however, a background in either cryptography or statistics or systems building will be very helpful, as the course content is at the intersection of these topics.
The course will be structured around paper readings, class discussions, and an individual research project.