BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:Representation Learning on Graphs - Jure Leskovec - Associate Prof
 essor of Computer Science at Stanford University\, Chief Scientist at Pint
 erest\, and investigator at Chan Zuckerberg Biohub
DTSTART:20190320T161500Z
DTEND:20190320T170000Z
UID:TALK120487@talks.cam.ac.uk
CONTACT:jo de bono
DESCRIPTION:Machine learning on graphs is an important and ubiquitous task
  with applications ranging from drug design to friendship recommendation i
 n social networks. The primary challenge in this domain is finding a way t
 o represent\, or encode\, graph structure so that it can be easily exploit
 ed by machine learning models. However\, traditionally machine learning ap
 proaches relied on user-defined heuristics to extract features encoding st
 ructural information about a graph. In this talk I will discuss methods th
 at automatically learn to encode graph structure into low-dimensional embe
 ddings\, using techniques based on deep learning and nonlinear dimensional
 ity reduction. I will provide a conceptual review of key advancements in t
 his area of representation learning on graphs\, including random-walk base
 d algorithms\, and graph convolutional networks. We will discuss applicati
 ons to web-scale recommender systems\, healthcare and knowledge representa
 tion and reasoning.\n\nThe slides from the talk: http://i.stanford.edu/~ju
 re/pub/talks2/graphsage_gin-cambridge-mar19.pdf\n\nThe talk was not record
 ed due to technical issues. \n
LOCATION:Lecture Theatre 2\, Computer Laboratory
END:VEVENT
END:VCALENDAR