BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:CANCELLED Probabilistic models of graphs for meaning representatio
 ns - Adam Lopez (University of Edinburgh)
DTSTART:20191119T100000Z
DTEND:20191119T110000Z
UID:TALK133483@talks.cam.ac.uk
CONTACT:Guy Aglionby
DESCRIPTION:Several datasets are now available that represent the meaning 
 of natural language sentences. The semantic theories underlying these data
 set often differ\, but they are alike in that the representations all take
  the form of graphs. So\, to properly model them in a machine learning sys
 tem we need probabilistic models of graphs. Unfortunately\, the space of g
 enerative probabilistic models of graphs is larger\, weirder\, and much le
 ss well explored than the space of generative models for strings. This tal
 k will give you a guided tour of this space\, including classical formalis
 ms based on automata and grammars\, and their more recent neural variants.
  It will cover work done in collaboration with Sorcha Gilroy\, Federico Fa
 ncellu\, Ieva Vasiljeva\, Pijus Simonaitis\, Sebastian Maneth\, and Mirell
 a Lapata.\n\nAdam Lopez is a Reader at the University of Edinburgh with ma
 ny research interests in natural language processing that are difficult to
  summarize.
LOCATION:FW26\, Computer Laboratory
END:VEVENT
END:VCALENDAR