Network Science for Multidimensional Data Analysis
The coexistence of multiple types of interactions within social, technological and biological networks has moved the focus of the statistical physics of complex systems towards their description as a set of subsystems organized as layers of connectivity. This novel approach has unveiled that the multilayer nature of complex systems has strong influence in the emergence of collective states and their critical properties.
Although recently spurred by the burst of datasets in which different means of interactions within the same system are encoded, the interest in the multilayered nature of complex systems dates back several decades and span across diverse disciplines. This has enhanced the activity devoted to multilayered networks of many network science practitioners during the recent years and, nowadays, the topic is one the most important directions in the field.
Session I: MIX-NEXT. Real-world systems are not only multilayer in their nature: they exhibit, simultaneously, a complex organization across multiple scales in their topology, dynamics and function. Recently, many approaches have been proposed to model higher-order interactions different from multiplexity: from simplicial complexes to memory in network flows, from latent topological geometry to multiresolution methods to unravel the hidden geometry of network-driven processes, from simplicial complexes to hypergraphs, and also the role of time evolution in all the mentioned structures. All those methods provide powerful tools to analyze complex systems and to unravel the effects of hierarchies from different points of view. However, empirical networks often exhibit multiscale spatio-temporal organization, multilayer relationships and non-trivial geometry.
The aim of this session is to balance the contribution of well established leading experts and rising young researchers to review the recent advances in those research fields, with the aim of triggering and igniting new discussions on theoretical and computational solutions required to build a more comprehensive set of tools integrating different perspectives into one, coherent and self-consistent, framework for modeling and analysis of complex networks.
Session II: State of the art. In this half-day session, we hope to explore the current state of the field of multilayer network science to characterize major challenges, identify important application areas, and key advances that we hope to make in the near future. The session will act as a kickoff meeting for AccelNet-MultiNet, a new NSF-sponsored initiative that aims to build new ties for collaboration, and expand the cohesiveness and coordination of the research community. The AccelNet-MultiNet program will include international scholar exchanges, collabathon activities, webinar series, and future in-person and virtual meetings that aim to formalize shared standards, priorities and strategies. Satellite participants will be invited to join the community to contribute and participate and shape future activities. Invited speakers include: Ginestra Bianconi, Byungnam Kahng, Sarika Jalan
Fondazione Bruno Kessler
Universitat Rovira i Virgili
Indian Institute of Tech. Indore
Seoul National University
U. Louvain and U. Namur
|1 July 2020||Deadline for speakers' early-bird registration|
|10 July 2020||Final Program|
|10 August 2020||Deadline for speakers' standard registration|
|18 September 2020||Satellite event: Session I|
|19 September 2020||Satellite event: Session II|
|18 Sep: 13:55||
Manlio De Domenico
Session I: Opening and Chair
Kwang-Il Goh, Korea University
Towards the signed network theory
Network science is undergoing another major transition of conceptual and technical maturation these days, following expansions through new "Beyond" areas. One promising area anticipating such development is the signed network theory. In this talk, some recent effort towards the signed network theory will be presented, which aims to go beyond the traditional structural balance-based approach and to establish a research framework for better understanding the structure of and the dynamics on signed networks. (Based on the works in collaboration with Sang-Hwan Gwak, Sungmin Lee, and Kyu-Min Lee.)
Clara Granell, Universitat Rovira i Virgili
Human response to epidemics: how selective prophylaxis can induce oscillations and how pulsating campaigns can palliate them
Human behavioral responses play an important role in the impact of disease outbreaks and yet they are often overlooked in epidemiological models. Understanding to what extent behavioral changes determine the outcome of spreading epidemics is essential to design effective intervention policies. Here we explore, analytically, the interplay between the personal decision to protect oneself from infection and the spreading of an epidemic. We do so by coupling a decision game based on the perceived risk of infection with a Susceptible-Infected-Susceptible model. Interestingly, we find that the simple decision on whether to protect oneself is enough to modify the course of the epidemics, by generating sustained steady oscillations in the prevalence. We deem these oscillations detrimental, and propose two intervention policies aimed at modifying behavioral patterns to help alleviate them. Surprisingly, we find that pulsating campaigns, compared to continuous ones, are more effective in diminishing such oscillations.
Dima Krioukov, Northeastern University
Ollivier-Ricci curvature convergence in random geometric graphs
Curvature is one of the most basic geometric characteristics of space. The original definitions of curvature apply only to smooth Riemannian or Lorentzian manifolds, but there exist numerous nonequivalent extensions of these definitions applicable to graphs and simplicial complexes. Unfortunately, no notion of graph curvature is known to converge in any limit to any curvature of any smooth space. We show that the Ollivier curvature of random geometric graphs in any Riemannian manifold converges in the continuum limit to the Ricci curvature of the manifold. Random geometric graphs are fundamental in topology since they are 1-skeletons of Rips complexes whose topology is known to converge to the manifold topology. We show that their geometry also converges to the manifold geometry. This result establishes the first rigorous connection between curvatures of random discrete objects and smooth continuous spaces, proving correct original Riemann's ideas behind the foundations of Riemannian geometry.
Michelle Feng, UCLA
Detecting and interpreting topological properties of real-word spatial systems
Tools from algebraic topology help characterize the global structure of mathematical spaces. In this talk, I will discuss methods for transforming real-world spatial systems into spaces which are amenable to topological analysis. I will present several methods for constructing a topological space from data (including the adjacency and level-set methods introduced by myself and collaborators), which I will compare using real-world voting data. I also discuss several other geospatial examples and explain what type of insights can be gleaned from topological information. In particular, I will focus on how to interpret topological properties of different data sets and different choices of topological space construction.
Leto Peel, U. catholique de Louvain / U. de Namur
Hierarchical community structure in networks
Modular and hierarchical structures are pervasive in real-world complex systems. A great deal of effort has gone into trying to detect and study these structures. Important theoretical advances in the detection of modular, or “community”, structures have included identifying fundamental limits of detectability by formally defining community structure using probabilistic generative models. Detecting hierarchical community structure introduces additional challenges alongside those inherited from community detection. In this talk I will present a theoretical study on hierarchical community structure in networks, which has thus far not received the same rigorous attention. We address the following questions: 1) How should we define a valid hierarchy of communities? 2) How should we determine if a hierarchical structure exists in a network? and 3) how can we detect hierarchical structure efficiently? We approach these questions by introducing a definition of hierarchy based on the concept of stochastic externally equitable partitions and their relation to probabilistic models such as the popular stochastic block model. We enumerate the challenges involved in detecting hierarchies and, by studying the spectral properties of hierarchical structure, present an efficient and principled method for detecting them. This is joint work with Michael Schaub.
Arsham Ghavasieh, Fondazione Bruno Kessler
Enhancing transport properties in interconnected systems without altering their structure
Units of complex systems –such as neurons in the brain or individuals in societies— must communicate efficiently to function properly: e.g., allowing electrochemical signals to travel quickly among functionally connected neuronal areas in the human brain, or allowing for fast navigation of humans and goods in complex transportation landscapes. The coexistence of different types of relationships among the units, entailing a multilayer representation in which types are considered as networks encoded by layers, plays an important role in the quality of information exchange among them. While altering the structure of such systems—e.g., by physically adding (or removing) units, connections, or layers—might be costly, coupling the dynamics of subset(s) of layers in a way that reduces the number of redundant diffusion pathways across the multilayer system, can potentially accelerate the overall information flow. To this aim, we introduce a framework for functional reducibility which allow us to enhance transport phenomena in multilayer systems by coupling layers together with respect to dynamics rather than structure. Mathematically, the optimal configuration is obtained by maximizing the deviation of system's entropy from the limit of free and noninteracting layers. Our results provide a transparent procedure to reduce diffusion time and optimize noncompact search processes in empirical multilayer systems, without the cost of altering the underlying structure.
Chiara Poletto, INSERM
Individual heterogeneities and multi-layer connectivity in the spread of epidemics
In the last decades, new network theories and epidemiological evidence have substantially advanced our knowledge about how acute respiratory infections spread in the human population. Temporal and multi-layer networks provide a paradigmatic example. These frameworks allow for clearly describing the heterogeneous connectivity features driving transmission, e.g. different connectivity by settings (household, workplace, school, etc.), and occasional vs. recurrent encounters. However, the role of individuals in transmission is determined, besides their connectivity, by their intrinsic properties, such as their susceptibility and their probability to develop a severe form of infection. Importantly, individuals' intrinsic properties and multi-layer connectivity are correlated which makes it difficult to disentangle their relative role on the epidemic unfolding and the effect of interventions. During the talk I will tackle this issue, presenting recent works applied to influenza and COVID-19 pandemic. I will discuss the role of children – who drive influenza transmission, but conversely seem to be protected by COVID-19 infection – and its implication for the spatiotemporal dynamics of epidemics and the impact of contact tracing.
Giovanni Petri, ISI
On the topological complexity of decision boundaries of neural networks
Complex data require complex models, or so the saying goes. However, the reason why classifying two concentric circles is challenging is not because they are circles, but because they are concentric. That is, not the properties of data themselves, but rather those of the task shape the complexity of a classification problem. We adopt here this perspective and, using simple neural networks and topological data analysis as explanatory tools, we attempt at shifting the focus from data-driven to task-oriented models. To this aim, I will discuss the topology of the decision boundaries of neural networks in simple classification problems as a measure of their intrinsic complexity. I will employ a topological approach to understand how the architecture of a model shapes its topological expressiveness and how complex data do not necessarily equate with difficult classification problems. Finally, I will discuss the insights that these observations provide on the topological expressiveness of neural network architectures. (Joint work with Antonio Leitao)
Open discussion: NEXT
|19 Sep: 15:30||
Ann McCranie and Santo Fortunato
Session II: Introduction
Adaptive multiplexing leading to explosive synchronization in multilayer networks
Link overlap influence opinion dynamics on multiplex networks: spin model approach
Multilayer networks: Structure and function
Panel discussion: Hub leaders and PIs