BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:Natural Conjugate Gradient Learning for Fixed-Form Variational Bay
 es - Dr Antti Honkela (Aalto University School of Science and Technology\,
  Finland)
DTSTART:20100616T100000Z
DTEND:20100616T110000Z
UID:TALK24175@talks.cam.ac.uk
CONTACT:Zoubin Ghahramani
DESCRIPTION:Variational Bayesian (VB) methods are typically only applied t
 o models in the conjugate-exponential family using the variational Bayesia
 n expectation maximisation (VB EM) algorithm or one of its variants. Here 
 I present an efficient algorithm for applying VB to more general models. T
 he method is based on specifying the functional form of the approximation\
 , such as multivariate Gaussian.  The parameters of the approximation are 
 optimised using a natural conjugate gradient algorithm that utilises the R
 iemannian geometry of the space of the approximations. This leads to a ver
 y efficient algorithm for suitably structured approximations. It is shown 
 empirically that the proposed method is comparable or superior in efficien
 cy to the VB EM in a case where both are applicable. The algorithm is also
  applied to learning a nonlinear state-space model and a nonlinear factor 
 analysis model for which the VB EM is not applicable. For these models\, t
 he proposed algorithm outperforms alternative gradient-based methods by a 
 significant margin.\n\nThis is joint work with Tapani Raiko\, Mikael Kuuse
 la\, Matti Tornio and Juha Karhunen.
LOCATION:Engineering Department\, CBL Room 438
END:VEVENT
END:VCALENDAR