BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:Approximate Bayesian Computation and Complex network models - Dami
 en Fay (Bournemouth University)
DTSTART:20140116T150000Z
DTEND:20140116T160000Z
UID:TALK48361@talks.cam.ac.uk
CONTACT:Eiko Yoneki
DESCRIPTION:This talk is broken into two sections. The first may be intere
 sting to those with complex processes and no clear means to estimate the p
 arameters of that process. The second looks specifically at complex networ
 k models.\n\nIn the general area of model fitting\, likelihood/Bayesian ba
 sed approaches are popular not just because of the excellent parameter est
 imates but also because they return the posterior distribution of the para
 meter – i.e. we can see how probable it is that a parameter actually has
  a different value from the one we are using. One can think for example of
  the case where the posterior is almost uniform which essentially means th
 e parameter estimate is useless\; in this case care should be taken when d
 eriving conclusions from the model.\n\nApproximate Bayesian Computation (A
 BC) is a method for estimating the posterior distribution for model parame
 ters when the a likelihood based approach doesn’t work\; typically this 
 occurs when the likelihood function a) doesn’t exist\, b) the likelihood
  is vanishingly small or c) it is just too complicated or time consuming t
 o derive (especially if one is not convinced the model is appropriate in t
 he first case). ABC has been around for many years but has recently become
  very popular due to some advances in the “summary statistic selection p
 roblem” – the Achilles heal of ABC .\n\nThe second half of the talk lo
 oks at graph topology generators. It may come as a surprise that 11 years 
 after the introduction of the preferential attachment model there still ex
 ists no likelihood expression for that model. For the small world model on
 e can only place bounds on the parameters. For models that include anythin
 g of greater complexity one is left without any option. This research thus
  asks the question\, can ABC be applied to any graph topology generator? W
 e then look at the answer and uncover some interesting technical issues le
 ading to a set of heuristics for using ABC . We also take some real world 
 data sets and show how using ABC we can select appropriate models for thos
 e datasets.\n\nBio: Damien Fay obtained a B.Eng from UCD (1995)\, an MEng 
 (1997) and PhD (2003) from DCU and worked as a mathematics lecturer at the
  National University of Ireland (2003-2007) before joining the NetOS group
 \, Computer Laboratory\, Cambridge from 2008-2010 as a research associate.
  He was a research fellow at the Cork centre for computational complexity\
 , 4C\, University College Cork from 2011-2012. Damien recently took up a t
 enured lectureship in big data analysis with the smart technologies resear
 ch centre at Bournemouth university. Damien is best known for developing t
 he wavelet transfer model in time series analysis\, the weighted spectral 
 distribution metric in applied graph theory and recently lodged a patent f
 or an arrival time prediction algorithm which has won a UCC commercialisat
 ion 2012 award. His interests are in the areas of time series analysis and
  applied graph theory.\n\n
LOCATION:FW26\, Computer Laboratory\, William Gates Builiding
END:VEVENT
END:VCALENDAR