Data Science Seminar: The Mathematics of Deception09 Mar 2022, by Sponsored events in
29 April 2022
Sidharth Jaggi, School of Mathematics, University of Bristol, UK
In a variety of information-processing tasks (say Alice wishes to store data, or transmit it to Bob, or estimate some underlying signal from some sensors), one has to design schemes that are robust to noise (servers may crash, or there may be noise in the transmission or sensing mechanisms). Fundamental performance limits, and algorithms attaining these fundamental limits, are relatively well-understood when this noise is random. In this talk we focus on scenarios where the noise is chosen by an adversary whose goal is to disrupt the information-processing task. In such settings, the fundamental limits are often more pessimistic than in the random noise setting, since the adversary can attempt to “spoof” the information being stored/transmitted/estimated. We will attempt to shed light on what is known in some adversarial noise settings, and showcase a few open problems.
In cooperation with the Jean Golding Institute, University of Bristol
More information and registration on the Bristol Data Science Seminar Series