BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:Title to be confirmedMaking sense of language: It's okay to count 
 - Gabe Recchia\, University of Cambridge
DTSTART:20150929T100000Z
DTEND:20150929T110000Z
UID:TALK61164@talks.cam.ac.uk
CONTACT:Microsoft Research Cambridge Talks Admins
DESCRIPTION:Vector-based representations have long been a popular choice f
 or researchers interested in unsupervised learning of word meaning. With t
 he 2013 release of word2vec\, the use of shallow neural language models (s
 ometimes called word embeddings or prediction-based models) for constructi
 ng such vectors has become extremely popular. In the past year\, however\,
  several researchers have demonstrated that traditional distributional mod
 els (sometimes called count-based models) are capable of similar levels of
  performance when properly parameterized. Furthermore\, count-based models
  give rise to more easily interpretable lexical representations\, making t
 hem preferable to neural models in certain use cases. I give examples of s
 everal areas in which simple co-occurrence based models demonstrate surpri
 singly high levels of utility or performance: predicting human judgments o
 f semantic relatedness and similarity\, estimation of geographic locations
 \, extraction of semantic topics\, and a toy question-answering task and d
 ataset recently proposed by researchers at Facebook AI Research as a possi
 ble step towards human-level natural language understanding. Applying a ve
 ry simple\, interpretable model to this dataset highlights benefits and sh
 ortcomings of the proposed task\, and points the way to improved training 
 and testing environments for natural language understanding systems.Abstra
 ct not available
LOCATION:Indigo 05-27\, Microsoft Research Ltd\, 21 Station Road\, Cambrid
 ge\, CB1 2FB
END:VEVENT
END:VCALENDAR