BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:[Special Statslab Seminar] Scalable methods for machine learning o
 ptimisation - Milan Vojnovic (LSE)
DTSTART:20190510T130000Z
DTEND:20190510T140000Z
UID:TALK120508@talks.cam.ac.uk
CONTACT:HoD Secretary\, DPMMS
DESCRIPTION:Optimisation methods play one of the most important roles in m
 achine learning area. High-dimensionality of machine learning models and l
 arge volume of training data introduce a variety of challenges\, both from
  the\nfundamental optimisation methodology perspective and distributed com
 putation perspective. In this talk\, I will present techniques that allow 
 us to accelerate training of machine learning models in distributed comput
 ing\nsystems\, and approximately solve certain classes of submodular optim
 isation problems by using simple surrogate functions. In both these proble
 ms\, we leverage combining lossy data compression with optimisation. Time 
 permitting\, I will also briefly discuss some recent results and open\nque
 stions that arise in online decision making under uncertainty\, statistica
 l relational learning\, and inverse problems for stochastic processes on g
 raphs.
LOCATION:MR12
END:VEVENT
END:VCALENDAR