Testing- Education & Outreach Courses/Open Course Material

CS 227r: Topics In Cryptography and Privacy

Semester: 

Fall

Offered: 

2014

Co-Instructors:  Kobbi Nissim & Or Sheffet

Meets: Tuesday/Thursday 11:30AM - 1:00PM in MD 119

Course Description: This course will cover topics in cryptography and data privacy drawn from the theoretical computer science research literature with particular focus on differential privacy -- a mathematical framework for privacy-preserving analysis of datasets, which enables aggregate computations while preventing the leakage of individual-level information.

Law Courses

Salil Vadhan

Salil Vadhan

Vicky Joseph Professor of Computer Science and Applied Mathematics, SEAS, Harvard
Area Chair for Computer Science

Salil Vadhan is the lead PI of Privacy Tools for Sharing Research Data project and the Vicky Joseph Professor of Computer Science and Applied Mathematics...

Read more about Salil Vadhan
Alexander Koujianos Goldberg. 3/2018. “Towards Differentially Private Inference on Network Data.” Applied Mathematics.Abstract
Statistical analysis of network data, while popular in a broad range of fields, can also be highly problematic from a privacy standpoint. In this thesis, we study privacy-preserving inference on network data using the rigorous notion of differential privacy. We propose new methods for differentially private inference using a common class of models known as Exponential Random Graph Models (ERGMs). The goal of our work is to accurately estimate the parameters of an ERGM applied to a network dataset, while offering meaningful privacy guarantees to participants. We propose methods that provably guarantee differential privacy at two different granularities: edge-level privacy, which protects the privacy of any single relationship in the network and node-level privacy, which protects all of the relationships of a participant. Specifically, using the framework of "restricted sensitivity," we take advantage of the sparsity of real-world networks to perturb data much less than prior work while guaranteeing differential privacy. We empirically evaluate the accuracy of inference in a series of experiments on both synthetic networks and a real network dataset. Experimental results suggest that our proposed methods enable accurate inference under meaningful privacy guarantees in settings where current methods do not, moving us closer to the goal of useful differentially private statistical modeling of network data.

Alexandra Wood and Stephen Chong to speak on "Robot Lawyers: Automating Legal Compliance for Transferring Private Data"

Privacy Tools team members Alexandra Wood and Stephen Chong are giving a talk on "Robot Lawyers: Automating Legal Compliance for Transferring Private Data." The talk will be on May 14th at 2:00pm at the Boston University School of Law. For more information, please visit https://www.bu.edu/law/news-events/events-calendar/?eid=209434.

Read more about Alexandra Wood and Stephen Chong to speak on "Robot Lawyers: Automating Legal Compliance for Transferring Private Data"

Alexander Goldberg Wins Hoopes Prize for Undergraduate Thesis: “Towards Differentially Private Inference on Network Data”

Alexander Goldberg, a graduating senior at Harvard University, has been awarded the Hoopes Prize for his undergraduate thesis, "Towards Differentially Private Inference on Network Data." This project was supervised and nominated by Privacy Tools Project Principal Investigator Salil Vadhan. Congratulations Alexander! Read more about Alexander Goldberg Wins Hoopes Prize for Undergraduate Thesis: “Towards Differentially Private Inference on Network Data”

Latanya Sweeney Holds Public Debate with French Secretary of State for Digital Commerce; Consults with President Macron on Tricolor Strategy on Artificial Intelligence

Privacy Tools Principal Investigator Latanya Sweeney held a public debate with the French Secretary of State for Digital Commerce about open data and AI algorithms. This debate was televised on French national television. Sweeney consulted with French President Emmanel Macron and joined him at the College de France as he presented the Tricolor Strategy on Artificial Intelligence. This groundbreaking public announcement revealed that there will be open data in France and that all AI algorithms used by the French government for decision-making will be made public, which...

Read more about Latanya Sweeney Holds Public Debate with French Secretary of State for Digital Commerce; Consults with President Macron on Tricolor Strategy on Artificial Intelligence