BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:Efficient Sequential Monte Carlo Inference for Kingman's Coalescen
 t - Dr Dilan Gorur (Gatsby Unit\, UCL)
DTSTART:20081022T120000Z
DTEND:20081022T130000Z
UID:TALK13783@talks.cam.ac.uk
CONTACT:Zoubin Ghahramani
DESCRIPTION:Algorithms for automatically discovering hierarchical structur
 e from data play an important role in machine learning. Teh et al. (2008) 
 proposed a Bayesian hierarchical clustering model based on Kingman's coale
 scent and proposed both greedy and sequential Monte Carlo (SMC) based aggl
 omerative clustering algorithms for inference\, the SMC algorithm having c
 omputational cost cubic in the number of data points per particle. \nWe bu
 ild upon this work and propose a new SMC based algorithm for inference in 
 the coalescent clustering model where the computations required to conside
 r merging each pair of clusters at each iteration is not discarded in subs
 equent iterations.  This improves the computational cost to be quadratic p
 er particle.  In experiments we show that our new algorithm achieves impro
 ved costs without sacrificing accuracy or reliability.
LOCATION:Engineering Department\, CBL Room 438
END:VEVENT
END:VCALENDAR