ELFT HÍRADÓ

[StatFizSzeminar <statfiz AT glu.elte.hu>]

Az ELTE Komplex Rendszerek Fizikája Tanszék következő teájára szóló 
meghívót minden kedves érdeklődő figyelmébe ajánljunk.
 

                             MEGHÍVÓ

        az ELTE Komplex Rendszerek Fizikája Tanszék teájára

                       Tiago P. Peixoto
                     (University of Bremen)

   Hierarchical Block Structures and High-Resolution Model Selection
                      in Large Networks


Many social, technological, and biological networks are composed of
modules, representing groups of nodes that are assumed to have a similar
role in the functioning of the network. The use of statistical
generative models that formally characterize this modular structure is a
powerful tool that allows the development of well-defined and principled
techniques to extract such information from empirical data. In this
talk, I present a hierarchical generative model which describes the
modular structure of networks at multiple scales. Using non-parametric
inference based on the minimum description length principle, I show that
the method is capable of separating signal from noise (i.e. does not
detect spurious modules) while at the same time possessing an increased
resolution, being able to detect very small modules in very large
networks. Furthermore, it fully generalizes other approaches in that it
is not restricted to purely assortative mixing patterns, directed or
undirected graphs, and ad hoc hierarchical structures such as binary
trees. Despite its general character, the approach is tractable, and
yields an efficient algorithm which scales well for very large networks.
I illustrate the application of the method on several empirical
networks.


Az előadás kezdete: szeptember 23-án, kedden kettőkor
            helye: ELTE TTK északi épület, 5.128-as terem.

Vendégeket, mint mindig, most is szívesen látunk.