BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:CNN seminar - January - Speaker to be confirmed
DTSTART:20120126T140000Z
DTEND:20120126T150000Z
UID:TALK35814@talks.cam.ac.uk
CONTACT:Petra Vertes
DESCRIPTION:*Vincenzo Nicosia* (Computer Laboratory\, University of Cambri
 dge\, UK and Laboratory on Complex Systems\, Scuola Superiore di Catania\,
  Italy):\n\n"Controlling centrality in complex networks"\n\nand\n\n*Eiko Y
 oneki* (Computer Laboratory\, University of Cambridge\, UK):\n\n"On Joint 
 Diagonalisation for Dynamic Network Analysis"\n\n*Abstracts:*\n\n*Controll
 ing centrality in complex networks*\n\nNode and edge centrality have a piv
 otal importance in the study and characterization of complex networks\, an
 d nowadays centrality measures are widely used to identify influential ind
 ividuals in social groups\, to rank Web pages by popularity\, and even to 
 determine the impact of scientific research. Many different structural pro
 perties have been used to assess the importance of nodes\, but in most of 
 the cases the centrality of every single node crucially depends on the ent
 ire pattern of connections. Therefore\, the usual approach is to compute n
 ode centralities once the network structure is assigned. We discuss here a
  solution to the so-called "inverse centrality problem"\, which consists i
 nto controlling the centrality scores of the nodes by opportunely acting o
 n the structure of a given network. In particular\, we focus our attention
  on spectral centrality measures and we show that there exist particular s
 ubsets of nodes\, called controlling sets\, which can assign any prescribe
 d set of centrality values to all the nodes of a graph\, by cooperatively 
 tuning the weights of their out-going links. We found that many large netw
 orks from the real world have surprisingly small controlling sets\, contai
 ning even less than 5-10% of the nodes. Consequently\, the rankings obtain
 ed by spectral centrality measures should be taken into account with extre
 me care\, since they can be easily manipulated and even distorted by small
  groups of malicious nodes acting cooperatively.\n\nReferences: [1] V. Nic
 osia\, R. Criado. M. Romance\, G. Russo and V. Latora "Controlling central
 ity in complex networks" Scientific Reports 2\, 218 (2012)\, doi:10.1038/s
 rep00218 http://www.nature.com/srep/2012/120111/srep00218/full/srep00218.h
 tml\n\n \n\n*On Joint Diagonalisation for Dynamic Network Analysis*\n\nJoi
 nt diagonalisation (JD) is a technique used to estimate an average eigensp
 ace of a set of matrices. Whilst it has been used successfully in many are
 as to track the evolution of systems via their eigenvectors\; its applicat
 ion in network analysis is novel. The key focus is the use of JD on matric
 es of spanning trees of a network. This is especially useful in the case o
 f real-world contact networks in which a single underlying static graph do
 es not exist. The average eigenspace may be used to construct a graph whic
 h represents the `average spanning tree' of the network or a representatio
 n of the most common propagation paths. We then examine the distribution o
 f deviations from the average and find that this distribution in real-worl
 d contact networks is multi-modal\; thus indicating several modes in the u
 nderlying network. These modes are identified and are found to correspond 
 to particular times. Thus JD may be used to decompose the behaviour\, in t
 ime\, of contact networks and produce average static graphs for each time.
  This may be viewed as a mixture between a dynamic and static graph approa
 ch to contact network analysis.\n\nhttp://arxiv.org/abs/1110.1198
LOCATION:Keynes Hall in Kings College
END:VEVENT
END:VCALENDAR