InfoSec Seminar: Adversarial Machine Learning

Speaker: Jamie Hayes

Date/Time: 18-Jan-2018, 16:00 UTC

Venue: New Quad Pop Up 102

Details

Abstract

This talk gives an overview of two recent papers on use cases of adversarial machine learning. The first half presents the first ever method for information hiding using machine learning that is competitive with more established techniques. The second half presents a case study into privacy leakage of generative models. With no knowledge of the training algorithm, model type or architecture, model parameters, or data distribution, we show an attacker can learn which data items were used to train a generative model, which is potentially privacy sensitive information.
 

Bio

Jamie Hayes is a PhD student from University College London, UK. His research focuses on the intersection of privacy, security and machine learning.

Add to Calendar

This page was last modified on 27 Mar 2014.