Please note: This PhD seminar will be given online.
Masoumeh
Shafieinejad, PhD
candidate
David
R.
Cheriton
School
of
Computer
Science
Supervisor: Professor Florian Kerschbaum
I will talk about my internship project at Microsoft Research during summer 2020. In this project, my collaborators and I investigate the limitations of differential privacy when used for confidentiality in social graphs. We focus on models that are trained on organizational communication data, e.g. emails. These models carry unique risks of breaching confidentiality, even if the model is intended only for internal use. Works that apply differential privacy techniques to natural language processing tasks usually assume independently distributed data, and overlook potential correlation among the records. Ignoring this correlation results in a fictional promise of privacy. Naively extending differential privacy techniques to focus on group privacy instead of record-level privacy is a straightforward approach to mitigate this issue. This approach, although providing a more realistic privacy-guarantee, is over-cautious and severely impacts model utility. We show this gap between these two extreme measures of privacy over two language tasks, and introduce a middle-ground solution. We propose a model that captures the correlation in the social network graph, and incorporates this correlation in the privacy calculations through Pufferfish privacy principles.
To join this PhD seminar on BigBlueButton, please go to https://bbb.crysp.org/b/mas-t2w-dkn.