BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:Fast Bayesian Boolean Matrix Factorisation - Chris Holmes (Univers
 ity of Oxford)
DTSTART:20170705T143000Z
DTEND:20170705T151500Z
UID:TALK73159@talks.cam.ac.uk
CONTACT:INI IT
DESCRIPTION:Boolean matrix factorisation decomposes a binary data matrix i
 nto an approximating Boolean product of two low rank\, binary matrices: on
 e containing meaningful patterns (signatures)\, the other quantifying how 
 the observations can be expressed as a logical combination of these patter
 ns.  &nbsp\;  <br><span><br>We introduce a probabilistic model for Boolean
  matrix factorisation\, termed the &ldquo\;OrMachine&rdquo\;\, and derive 
 a Metropolised Gibbs sampler that facilitates efficient parallel posterior
  inference on commodity hardware. On real world and simulated data\, our B
 ayesian method provides state of the art performance for Boolean matrix fa
 ctorisation and matrix completion. The method supports full posterior infe
 rence\, which is important in applications\, for example in controlling fa
 lse positive rates in collaborative filtering and\, crucially\, improves t
 he interpretability of the inferred patterns. The proposed model and compu
 tation scale to large datasets as motivated by an analysis of single cell 
 gene expression data recording measurements from 1.3 million mouse brain c
 ells across 11 thousand genes.&nbsp\;</span>
LOCATION:Seminar Room 1\, Newton Institute
END:VEVENT
END:VCALENDAR