BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//132.216.98.100//NONSGML kigkonsult.se iCalcreator 2.20.4//
BEGIN:VEVENT
UID:20250812T154230EDT-7030tVn8Lr@132.216.98.100
DTSTAMP:20250812T194230Z
DESCRIPTION:Abstract: Graph Neural Network (GNN) with its ability to integr
 ate graph information has been widely used for data analyses. However\, th
 e expressive power of GNN has only been studied for graph-level tasks but 
 not for node-level tasks\, such as node classification\, where one tries t
 o interpolate missing nodal labels from the observed ones. In this paper\,
  we study the expressive power of GNN for the said classification task\, w
 hich is in essence a function interpolation problem. Explicitly\, we deriv
 e the number of weights and layers needed for a GNN to interpolate a band-
 limited function in . Our result shows that\, the number of weights needed
  to -approximate a bandlimited function using the GNN architecture is much
  fewer than the best known one using a fully connected neural network (NN)
  - in particular\, one only needs weights using a GNN trained by samples t
 o -approximate a discretized bandlimited signal in . The result is obtaine
 d by drawing a connection between the GNN structure and the classical samp
 ling theorems\, making our work the first attempt in this direction.\n
DTSTART:20230130T213000Z
DTEND:20230130T223000Z
LOCATION:Room 708\, Burnside Hall\, CA\, QC\, Montreal\, H3A 0B9\, 805 rue 
 Sherbrooke Ouest
SUMMARY:Martina Neuman (Michigan State University)
URL:/mathstat/channels/event/martina-neuman-michigan-s
 tate-university-345116
END:VEVENT
END:VCALENDAR