BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Memento EPFL//
BEGIN:VEVENT
SUMMARY:Latent space models for multiplex networks with shared structure
DTSTART:20211029T151500
DTEND:20211029T170000
DTSTAMP:20260414T211934Z
UID:91a72b21d0e04c3549dc5ba07bcad96de8e94f385d68fde9dd5961f9
CATEGORIES:Conferences - Seminars
DESCRIPTION:Elizaveta (Liza) Levina\, University of Michigan\nLatent space
  models are frequently used for modelling single-layer networks and includ
 e many popular special cases\, such as the stochastic block model and the 
 random dot product graph.   Yet in practice\, more complex network struc
 tures are becoming increasingly common.  Here we propose a new latent spa
 ce model for multiplex networks: multiple\, heterogeneous networks observe
 d on a shared node set. Multiplex networks can represent a network sample 
 with shared node labels\, a network evolving over time\, or a single netwo
 rk with multiple types of edges.\nThe key feature of our model is that it 
 learns from data how much of the network structure is shared between layer
 s\, and pools information across layers as appropriate. We establish ident
 ifiability\, develop a fitting procedure using convex optimization in comb
 ination with a nuclear norm penalty\, and prove a guarantee of recovery fo
 r the latent positions as long as there is sufficient separation between t
 he shared and the individual latent subspaces. \nThe new model compares f
 avorably to other methods in the literature on simulated networks and on a
  multiplex network describing the worldwide trade of agricultural products
 . \n\nJoint work with Peter MacDonald and Ji Zhu.  \n 
LOCATION:https://epfl.zoom.us/j/65289358081
STATUS:CONFIRMED
END:VEVENT
END:VCALENDAR