CS Forum: Ivy K. Y. Woo "Ultima Thule: Journey from Actuarial Science to Decoy Sampling (and beyond)"
Time: Tuesday 21 Sep at 11:00-11:40
Place: Lecture room T2 in CS building
You can also attend the talk via Zoom https://aalto.zoom.us/j/62426151190 (passcode 117242)
Ultima Thule: Journey from Actuarial Science to Decoy Sampling (and beyond)
Ivy K. Y. Woo
Actuarial science and decoy sampling? These are terms that one will not see next to each other everyday. However, they are important milestones of my bumpy journey towards computer science, which I will walk you through in this talk.
In the first part of the talk, I will introduce actuarial science -- the study of insurance mathematics. This is a unique discipline that lies in the intersection of statistics, financial mathematics, economics and risk management. For the interest of the computer science audience, I will also discuss applications of data science to actuarial science.
The second part of the talk is on my first research project in computer science -- how graph-based deanonymisation attacks on ring signatures can be defeated. In a ring signature scheme, in attempt to achieve anonymity, a signer issues a signature on behalf of a "ring", which is a set consisting of the signer herself and a number of decoy users sampled by an external ring sampler. Since the ring size is typically chosen to be much smaller than the total number of users for efficiency, it was unclear whether an adversary could, given sufficient computing power, deanonymise signers solely by graph analysis. I will present evidences showing that, by setting the ring size to be logarithmic in the total number of users, a graph-analysing adversary can perform no better than random guessing.
Ivy K. Y. Woo obtained her MSc in Finance (Actuarial Science) from Ulm University and her BSc in Actuarial Science from The University of Hong Kong. She is currently on the way of a career change and exploring different areas of computer science. At the moment, she is looking into transportation networks and machine learning for image manipulations, among other topics.
Professor Chris Brzuska, Department of Computer Science