BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY: Herding: Driving Deterministic Dynamics To Learn And Sample Proba
 bilistic Models - Yutian Chen (University of California at Irvine)
DTSTART:20131031T150000Z
DTEND:20131031T163000Z
UID:TALK48589@talks.cam.ac.uk
CONTACT:Konstantina Palla
DESCRIPTION:\nThe herding algorithm was proposed as a deterministic dynami
 c system that integrated learning and inference for discrete Markov Random
  Fields. It mitigates the slow mixing problem of regular learning algorith
 ms based on MCMC methods. Also\, the pseudo-samples generated by herding e
 njoy a fast convergence rate on the average sufficient statistics to the e
 mpirical values in the training data set. \n\nIn this talk\, I will first 
 introduce the basic idea of the herding algorithm and talk about a few dis
 tinguished properties compared to standard learning algorithms for MRFs. T
 hen I would like to show two extensions of herding as pure deterministic s
 ampling algorithms that apply to discrete and continuous state space respe
 ctively. The fast convergence property is preserved in these extensions un
 der proper conditions. If time permitted\, I will also briefly introduce a
 nother application of herding to structured prediction problems including 
 image segmentation and the Go game prediction.
LOCATION:Engineering Department\, CBL Room 438
END:VEVENT
END:VCALENDAR