Speaker
Prof. Manlio de Domenico (Fundazione Bruno Kessler - ICT)

Description

Per què les notícies falses s'estenen més i més ràpid que les notícies reals? Quins són els bots socials i quin és el seu paper en la divulgació de la informació errònia? En aquesta xerrada respondrem aquest tipus de preguntes i mostrarem com és possible manipular l'opinió pública amb la nova ciència de les xarxes i la intel·ligència artificial. Com a cas pràctic, comentarem els fets ocorreguts en les xarxes socials durant el referèndum català de l'1 d'octubre de 2017. A la part final de la nostra xerrada, proporcionarem algunes pautes i contramesures per defensar-nos de la desinformació i la manipulació en línia.

Speaker
Joaquín Goñi (Purdue University, USA)

Description

In the 17th century, physician Marcello Malpighi observed the existence of patterns of ridges and sweat glands on fingertips. This was a major breakthrough and originated a long and continuing quest for ways to uniquely identify individuals based on fingerprints. In the modern era, the concept of fingerprinting has expanded to other sources of data, such as voice recognition and retinal scans. It is only in the last few years that technologies and methodologies have achieved high-quality data for individual human brain imaging, and the subsequent estimation of structural and functional connectivity. In this context, the next challenge for human identifiability is posed on brain data, particularly on brain networks, both structural and functional.

Here I present how the individual fingerprint of a human structural or functional connectome (as represented by a network) can be maximized from a reconstruction procedure based on group-wise decomposition in a finite number of orthogonal brain connectivity modes. By using data from the Human Connectome Project and from a local cohort, I also introduce different extensions of this work, including an extended version of the framework for inter-scanner identifiability, and an extended version of the framework for disentangling heritability and environmental brain network traits.

Speaker
Jordi Piñero (UPF)

Description

In this talk we will discuss the concept of ``liquid brains'' as the widespread class of cognitive living neural networks characterised by a common feature: the agents (ants or immune cells, for example) move in space. Thus, no fixed, long-term agent-agent connections are maintained. This stands in contrast with standard neural systems. How such a class of systems are capable of displaying cognitive abilities, from learning to decision-making? Collective dynamics, memory and learning properties of liquid brains is explored under the perspective of statistical physics. 

Using a comparative approach, we review the generic properties of three large classes of systems, namely: standard neural networks (``solid brains''), ant colonies and the immune system. We show that, in spite of their idiosyncratic differences, these systems do share key statistical properties with standard neural systems in terms of formal descriptions, while strongly depart in other ways. On one hand, the attractors found in liquid brains are not always based on connection weights but instead on population abundances. Moreover, some liquid systems use fluctuations in ways similar to those found in cortical networks, suggesting a relevant role of criticality as a way of rapidly reacting and adapting to external signals. Finally, we will also outline the computational and evolutionary aspects for the immune system as a liquid brain and its implications on the network structure and dynamics.

Speaker
Ernesto Estrada (Institute of Applied Mathematics, Universidad de Zaragoza)

Description

I will motivate the problem of studying long-range interactions in discrete complex systems, illustrated by some experimental results on the diffusion of adatoms and admolecules on metallic surfaces. I will speculate about other discrete complex systems where such effects can also be observed. Then, I will introduce the d-path Laplacian operators as a natural way to model such systems. I will prove some analytical results about the boundedness and self-adjointness of these operators. Then, I will introduce a generalization of the diffusion equation that takes into account such long-range effects. I will prove that under certain specific transformations of the d-path Laplacians we can reproduce the superdiffusive behaviour observed experimentally. I will clarify the differences between this model and the "random walks with Levy flights" as well as with the use of fractional calculus. I will give some snapshots of extensions to synchronization, epidemic spreading studies and nonlinear diffusion models.
Finally, I will introduce the concept of "metaplexes" in which we combine the internal structure of nodes, modelled as a continuous or discrete space, coupled with the discrete structure of inter-nodal connections. I will show some results about how the internal structure of nodes influences the global dynamics of a metaplex and some potential areas for extension.

 

Speaker
Stefanella Boatto (IM, Universidade Federal de Rio de Janeiro, Brazil)

Description

Migratory fluxes of humans and of insects of various species have favoured the spreading of diseases world-wise. In particular the Ae. Agypti and Ae. Albopcitus mosquitoes of the Aedes family, are vectors able to transmit and spread among humans a variety of diseases : Dengue, Zika, Chikungunya, Yellow fever and, the newly discovered, Mayaro (Hotez et al. PLoS Negl. Trop Dis. 2017). The Ae. Albopictcus, able to survive even at low temperature, is already well established in Europe, while the Ae. Aegypti, traditionally present in tropical regions are now starting colonizing part of Europe. The overlapping of the two mosquitoes is worrisome since it could increase the spreading of the concerned diseases. In France recent cases of locally transmitted Chikungunya have been reported (22 August, 2017) in addition to locally transmitted cases of Dengue virus type 1 (DENV-1) already registered in Nimes, south of France, in 2015. Dengue is rather invasive epidemic due to the fact that already four different serotypes are present. It is important to stress that those epidemics can have strong social and economical impacts if not seriously controlled. Only in 2010 in Brazil, one million infected individual of which 80,000 where hospitalized. 

I shall revisite the SIR model with birth and death terms and time-varying infectivity parameter β(t) and introduce a network extension of it, SIR.Network. For a quite general slowly varying β(t) (not necessarily periodic) infectivity parameter we prove the existence of an attractor and we are able to determine an approximation : all the trajectories of the system are proven to be attracted into a tubular region around a suitable curve, which is an approximation of the underlying attractor. Numerical simulations are given and data fitting with real data from Dengue epidemics in Rio de Janeiro and So Paulo allow us to estimate the infectivity rate and make predictions about what are the periods more at risk of infection. A possible epidemic attractor is visualized and approximated. Finally I shall talk about work in progress with data from all over Brasil.