[Seminar] "Topological Communities in Complex Networks" by Luis Seoane

Title: Topological Communities in Complex Networks

 

Abstract: You have a complex network that summarizes your object of study: a neural or gene circuit, species in an ecosystem, people in a society, proteins in a metabolic pathway. What should you do with it? What kind of analysis will tell you what features you should pay attention to?

A first strategy is to eyeball the network—just “see” how it looks like, what stands out, etc. But networks are great tangles. The number of potential features, such as nodes forming communities, grow combinatorially. Key facets might not show up in a plot. Many algorithms automatically extract Geometric Communities—compact building blocks arising from nodes more densely connected between themselves than to others. What about aspects not defined by proximity? How to detect bridges between pairs of dense communities? What about subtle backbones of disperse, weaklyconnected nodes that hold a graph together?

In this talk I introduce Topological Communities: A framework and algorithm by which any network guides us automatically, in order of importance, to its most salient features. Critically, these features are not defined by node proximity, but by similarity of their topological roles. We thus detect relevant properties shared by nodes located far apart—inaccessible to existing methods. We illustrate our methods on the world airport graph, collaborations between representatives in the US House, human connectomes, etc.

Topological Communities decompose networks into alternative building blocks. They are great to try on a graph that we know nothing about. They let networks guide our attention towards their relevant aspects, offering a way in to start disentangling them. As part of my TSVP visit, I call for people across OIST who might have networks that might be interesting to look under this lens. Come by and let’s discuss Topological Communities from your data!
 

Speaker's short bio: Luis F Seoane heads the BIT-Lab at the Institute for Evolutionary Biology, of the Spanish National Research Council. His background is in Physics & Computational Neuroscience, with a PhD in Complex Systems. He has works in neurosicence, maths of Darwinian evolution, complex networks, linguistics, or econophysics. He often studies how complexity affect systems (e.g. how it makes them grow, thrive, or fail) and vice versa (e.g. how a system’s size or success might alter its complexity). Personal Homepage

 

This talk will also be broadcast online via Zoom:
Meeting ID: 993 1216 5065
Passcode: 603487