Class Time: Thursdays 12:10-2:00pm
Class Location: CSB 480


Data scientists and engineers have significant ethical and legal responsibilities to protect the privacy and best interests of the customers whose data they collect and use. This class discusses these responsibilities, the challenges of meeting them in practice, and a set of advanced technologies that can be used to enhance privacy, accountability, and data protection in big data systems. The focus (and uniqueness) of the class is to look at these technologies with a systems perspective of incorporating them in real data infrastructure systems. The class will cover advanced privacy technologies such as differential privacy, homomorphic databases, secure multi-party computation, hardware enclaves, and private federated learning.

For each of these technologies, our educational goal is two-fold: we both provide students with an understanding of the theoretical underpinnings of that technology and with working knowledge of where and how that technology can be applied in practice, along with the challenges and tradeoffs that may arise from such deployments. To this end, we organize our discussion of each technology in two stages: first, instructors introduce and demonstrate the basic theoretical concepts in a lecture-style manner; then, we all read technical materals (such as scientific papers or documentation) related to the technology and discuss them in class. We bias our selection of reading materials toward those that discuss practical application of the technologies at large institutions, such as Google, Apple, Meta, and governmental agencies.

Assignments in this class are also organized based on our two-fold educational goal. In the first half of the semester, while students accumulate basic understanding of privacy technology, we assign a set of homeworks that help students gain some basic working experience with each technology. The homeworks are narrow in scope, as they are individually solved and due every two weeks. In the second half of the semester, students swich to working on a group project, which they select on their own (from a list of admissible templates). The projects give students a chance to apply the concepts they’ve learned in a broader, more ambitious setting that (hopefully) solves a real problem. Before switching from homeworks to projects, students take a midterm quiz in class. There is no final exam, but there will be a final project presentation in the timeslot assigned by the registrar for the exam.


Grading


Textbook


Prerequisites:

  1. COMS W3137 Data Structures and Algorithms
  2. COMS W3157 Advanced Programming
  3. (Optional but useful to gain more from this class) Security I