BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:Structured deep models: Deep learning on graphs and beyond - Thoma
 s Kipf (Uni of Amsterdam)
DTSTART:20180621T100000Z
DTEND:20180621T110000Z
UID:TALK106945@talks.cam.ac.uk
CONTACT:39846
DESCRIPTION:In the recent years there has been an increasing number of suc
 cess stories in applying deep learning techniques to graph-structured data
 . The workhorse in this emerging field is the graph neural network: a mess
 age passing algorithm parameterized by neural networks\, trained via backp
 ropagation. Variants of graph neural networks now define the state of the 
 art in many classical graph or network problems\, such as node classificat
 ion\, graph classification\, and link prediction. \n\nIn this talk\, I wil
 l give an overview of structured deep models that employ graph neural netw
 orks as a key component and discuss trade-offs for a few popular model var
 iants such as graph convolutional networks (GCNs) [1] and graph attention 
 networks (GATs) [2]. I will further introduce two emerging research direct
 ions: learning deep generative models of graphs and inference of latent gr
 aph structure. Structured deep models are ideal candidates for these areas
  and hold promise for applications such as chemical synthesis\, program in
 duction\, and modeling of interacting physical and multi-agent systems.\n\
 n[1] Kipf & Welling\, Semi-supervised classification with graph convolutio
 nal networks\, ICLR 2017\n[2] Veličković et al.\, Graph attention networ
 ks\, ICLR 2018
LOCATION:Engineering Department\, CBL Room BE-438.
END:VEVENT
END:VCALENDAR