BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Memento EPFL//
BEGIN:VEVENT
SUMMARY:Graph learning: perspectives in statistics and signal processing
DTSTART:20160805T103000
DTEND:20160805T123000
DTSTAMP:20260407T011335Z
UID:397ea5aafcd3d3666c1193b52ab1bae095e34af2700121e4bb50ac2b
CATEGORIES:Conferences - Seminars
DESCRIPTION:Rodrigo Cerqueira Gonzalez Pena\nEDIC Candidacy Exam\nExam Pre
 sident: Prof. Jean-Philippe Thiran\nThesis Director: Prof. Pierre Vandergh
 eynst\nCo-examiner: Prof. Dimitri Van De Ville\nBackground papers:Model se
 lection through sparse maximum likelihood estimation for multivariate Gaus
 sian or binary data\, by O. Banerjee\, L. El Ghaoui\, and A. d’Aspremont
 .How to learn a graph from smooth signals\, by V. Kalofolias.Compressive s
 pectral clustering\, by N. Tremblay\, G. Puy\, R. Gribonval\, P. Vanderghe
 ynstAbstract:\nIt is increasingly common for machine learning algorithms t
 o take advantage of the relationship between data points to improve on the
 ir classification or regression performance. This data relationship can be
  efficiently modelled as a graph\, and it is known that the performance of
  algorithms such as spectral clustering\, or label propagation depend sole
 ly on the quality of this graph. Nonetheless\, if this graphical structure
  is not know in advance\, then it should be learned from the data observat
 ions. We present two approaches to solving this problem: the first learns 
 a graph in which the signals given as input are smooth according to some o
 bjective measure\; the second learns a graphical representation in the con
 text of a Gauss-Markov Random Field (GMRF) model\, by maximising the likel
 ihood of the model conditional on the observed data. We finally present a 
 breadth paper\, highlighting the importance of the learned graph for the p
 roblem of compressive spectral clustering.
LOCATION:ELE 242 http://plan.epfl.ch/?lang=en&room=ELE+242
STATUS:CONFIRMED
END:VEVENT
END:VCALENDAR