BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//132.216.98.100//NONSGML kigkonsult.se iCalcreator 2.20.4//
BEGIN:VEVENT
UID:20250802T185921EDT-7220dlXMvP@132.216.98.100
DTSTAMP:20250802T225921Z
DESCRIPTION:Title: Graph Attention Retrospective.\n\nGraph-based learning i
 s a rapidly growing sub-field of machine learning with applications in soc
 ial networks\, citation networks\, and bioinformatics. One of the most pop
 ular type of models is graph attention networks. These models were introdu
 ced to allow a node to aggregate information from the features of neighbor
  nodes in a non-uniform way in contrast to simple graph convolution which 
 does not distinguish the neighbors of a node. In this paper\, we study the
 oretically this expected behaviour of graph attention networks. We prove m
 ultiple results on the performance of the graph attention mechanism for th
 e problem of node classification for a contextual stochastic block model. 
 Here the features of the nodes are obtained from a mixture of Gaussians an
 d the edges from a stochastic block model where the features and the edges
  are coupled in a natural way. First\, we show that in an 'easy' regime\, 
 where the distance between the means of the Gaussians is large enough\, gr
 aph attention maintains the weights of intra-class edges and significantly
  reduces the weights of the inter-class edges. As a corollary\, we show th
 at this implies perfect node classification independent of the weights of 
 inter-class edges. However\, a classical argument shows that in the 'easy'
  regime\, the graph is not needed at all to classify the data with high pr
 obability. In the 'hard' regime\, we show that every attention mechanism f
 ails to distinguish intra-class from inter-class edges. We evaluate our th
 eoretical results on synthetic and real-world data.\n\nhttps://us06web.zoo
 m.us/j/85327310903?pwd=SlhEak53S2xrNkVYKzl4YUd5KzBudz09\n\nSite web : http
 s://dms.umontreal.ca/~mathapp/\n
DTSTART:20221128T210000Z
DTEND:20221128T220000Z
SUMMARY:Kimon Fountoulakis (University of Waterloo)
URL:/mathstat/channels/event/kimon-fountoulakis-univer
 sity-waterloo-343764
END:VEVENT
END:VCALENDAR