BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:Modelling Large- and Small-Scale Brain Networks - Thomas Nichols (
 University of Warwick)
DTSTART:20160714T130000Z
DTEND:20160714T133000Z
UID:TALK66756@talks.cam.ac.uk
CONTACT:INI IT
DESCRIPTION:Investigations of the human brain with neuroimaging have recen
 tly seen a  dramatic shift in focus\, from "brain mapping"\, identifying b
 rain regions related  to particular functions\, to connectivity or "connec
 tomics"\, identifying networks  of coordinated brain regions\, and how the
 se networks behave at rest and during  tasks. In this presentation I will 
 discuss two quite different approaches to  modeling brain connectivity. In
  the first work\, we use Bayesian time series  methods to allow for time-v
 arying connectivity. Non-stationarity connectivity  methods typically use 
 a moving-window approach\, while this method poses a single  generative mo
 del for all nodes\, all time points. Known as a "Multiregression  Dynamic 
 Model" (MDM)\, it comprises an extension of a traditional Bayesian  Networ
 k (or Graphical Model)\, by posing latent time-varying coefficients that  
 implement a regression a given node on its parent nodes. Intended for a mo
 dest  number of nodes (up to about 12)\, a MDM allows inference of the str
 ucture of the  graph using closed form Bayes factors (conditional on a sin
 gle estimated  "discount factor"\, reflecting the balance of observation a
 nd latent variance.  While originally developed for directed acyclic graph
 s\, it can also accommodate  directed (possibly cyclic) graphs as well. In
  the second work\, we use mixtures  of simple binary random graph models t
 o account for complex structure in brain  networks. In this approach\, the
  network is reduced to a binary adjacency matrix.  While this is invariabl
 y represents a loss of information\, it avoids a  Gaussianity assumption a
 nd allows the use of much larger graphs\, e.g. with 100&#39\;s  of nodes. 
 Daudin et al. (2008) proposed a "Erdos-Reyni Mixture Model"\, which  assum
 es that\, after an unknown number of latent node classes have been  estima
 ted\, that connections arise as Bernoulli counts\, homogeneously for each 
  pair of classes. We extend this work to account for multisubject data (wh
 ere  edge data are now Binomially distributed)\, allowing <br><br>Related 
 Links <ul> <li><a target="_blank" rel="nofollow">http://warwick.ac.uk/teni
 chols</a>  - Home page of Prof. Nichols&nbsp\;</li></ul>
LOCATION:Seminar Room 1\, Newton Institute
END:VEVENT
END:VCALENDAR