University of Birmingham > Talks@bham > Computer Security Seminars > Private machine learning: case of the Internet

Private machine learning: case of the Internet

Add to your list(s) Download to your calendar using vCal

If you have a question about this talk, please contact Vincent Cheval.

In this talk, I will briefly discuss of applying machine learning techniques for analysing ISP data. The main challenges are that data-sharing is poor and conventional private computation techniques are not scalable. I will describe a few privacy-preserving techniques that achieve a range of primitives such as vector division and normalisation—necessary for implementing machine learning techniques over large traffic matrices—which are implemented as distributed security protocols over partial homomorphic crytosystems. We overcome performance bottlenecks and give fast results for data produced by global-scale architectures. However while we make (yet another) case for ISP cooperation, I see vast scope for improvement in these protocols such as the use of formal verification techniques to achieve rigourous security guarantees.

This talk is part of the Computer Security Seminars series.

Tell a friend about this talk:

This talk is included in these lists:

Note that ex-directory lists are not shown.


Talks@bham, University of Birmingham. Contact Us | Help and Documentation | Privacy and Publicity.
talks@bham is based on from the University of Cambridge.