BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:Stream-based Statistical Machine Translation - Abby Levenberg\, Un
 iversity of Oxford
DTSTART:20111118T120000Z
DTEND:20111118T130000Z
UID:TALK34548@talks.cam.ac.uk
CONTACT:Ekaterina Kochmar
DESCRIPTION:We investigate a new approach for  SMT system training within 
 the *streaming* model of computation. We develop and test incrementally re
 trainable models which\, given an incoming stream of new data\, can effici
 ently incorporate the stream data online. A naive approach using a stream 
 would use an unbounded amount of space. Instead\, our online SMT system ca
 n incorporate information from unbounded incoming streams and maintain con
 stant space and time. Crucially\, we are able to match (or even exceed) tr
 anslation performance of comparable systems which are batch retrained and 
 use unbounded space. Our approach is particularly suited for situations wh
 en there is arbitrarily large amounts of new training material and we wish
  to incorporate it efficiently and in small space.  Our stream-based SMT s
 ystem is efficient for tackling  massive volumes of new training data and 
 offers-up new ways of thinking about translating web data and dealing with
  other natural language streams.\n
LOCATION:SS03\, Computer Laboratory
END:VEVENT
END:VCALENDAR