BIDSA SEMINAR: Learning & Privacy in Online Search
Douglas Leith , Trinity College Dublin
April 23, 2018
Bocconi University, 12:30pm
Via Rӧntgen n. 1, Rm 2 E4 SR 03
Abstract
From buying books to finding the perfect partner, we share our wants and needs
with our favourite online systems. But how far should we accept promises of
privacy in the face of personal profiling? In particular we ask how can we
improve detection of sensitive topic profiling by online search systems? We
propose a definition of privacy disclosure we call
{\epsilon}-indistinguishability from which we construct scalable, practical
tools to assess an adversaries learning potential. We demonstrate our results
using openly available resources, detecting a learning rate in excess of 98%
for a range of sensitive topics during our experiments. Using these detection
tools we also evaluate a range of possible approaches for disrupting learning.
Based on the following papers:
https://arxiv.org/abs/1504.08043
https://arxiv.org/abs/1703.03471
https://arxiv.org/abs/1509.05789