|COOKIES: By using this website you agree that we can place Google Analytics Cookies on your device for performance monitoring.|
Exclusive Pólya Urns and their applications
If you have a question about this talk, please contact Microsoft Research Cambridge Talks Admins.
o This event may be recorded and made available internally or externally via http://research.microsoft.com. Microsoft will own the copyright of any recordings made. If you do not wish to have your image/voice recorded please consider this before attendin
The Dirichlet Process (DP) and its variants have many nice mathematical properties which make them popular tools in the machine learning community. However, deploying DPs can be costly in settings where approximate inference over latent states is needed (e.g. Gibbs sampling seating arrangements in a CRP ).
By dropping some of the “nice” mathematical properties of the DP, we can construct a novel kind of stochastic process which offers exact inference, is fast, easy to implement, and can in many cases be used as a drop-in replacement for Chinese Restaurant Processes and Pitman-Yor processes. I will show how the new process differs from models in the existing literature, and how it can be applied to interesting tasks, including hierarchical sequence modelling and data compression.
This talk is part of the Microsoft Research Machine Learning and Perception Seminars series.
This talk is included in these lists:
Note that ex-directory lists are not shown.
Other listsHorizon Forum: The Cell-Materials Interface Centre for Family Research Seminar Series CRASSH lectures
Other talksCoal, steam and ships: economic historians versus historians of technology? Impingement of a plume on a rigid boundary Morphogenetic mechanisms underlying germinal and somatic fate acquisition in maize anthers PLACE Book Launch Feudalism in the Medieval West: ‘The Tyranny of a Construct’ Endodermal cell responses and size control regulate lateral root formation