Showing papers by "Vito Latora published in 2021"

Evolutionary dynamics of higher-order interactions in social networks.

Abstract: We live and cooperate in networks. However, links in networks only allow for pairwise interactions, thus making the framework suitable for dyadic games, but not for games that are played in larger groups. Here, we study the evolutionary dynamics of a public goods game in social systems with higher-order interactions. First, we show that the game on uniform hypergraphs corresponds to the replicator dynamics in the well-mixed limit, providing a formal theoretical foundation to study cooperation in networked groups. Second, we unveil how the presence of hubs and the coexistence of interactions in groups of different sizes affects the evolution of cooperation. Finally, we apply the proposed framework to extract the actual dependence of the synergy factor on the size of a group from real-world collaboration data in science and technology. Our work provides a way to implement informed actions to boost cooperation in social groups. Alvarez-Rodriguez et al. examine group interactions by means of higher-order social networks. They propose a theoretical framework for studying real-world interactions and provide a case study of collaboration in science and technology.

The physics of higher-order interactions in complex systems

Abstract: Complex networks have become the main paradigm for modelling the dynamics of interacting systems. However, networks are intrinsically limited to describing pairwise interactions, whereas real-world systems are often characterized by higher-order interactions involving groups of three or more units. Higher-order structures, such as hypergraphs and simplicial complexes, are therefore a better tool to map the real organization of many social, biological and man-made systems. Here, we highlight recent evidence of collective behaviours induced by higher-order interactions, and we outline three key challenges for the physics of higher-order systems. Network representations of complex systems are limited to pairwise interactions, but real-world systems often involve higher-order interactions. This Perspective looks at the new physics emerging from attempts to characterize these interactions.

Stability of synchronization in simplicial complexes.

Abstract: Various systems in physics, biology, social sciences and engineering have been successfully modeled as networks of coupled dynamical systems, where the links describe pairwise interactions. This is, however, too strong a limitation, as recent studies have revealed that higher-order many-body interactions are present in social groups, ecosystems and in the human brain, and they actually affect the emergent dynamics of all these systems. Here, we introduce a general framework to study coupled dynamical systems accounting for the precise microscopic structure of their interactions at any possible order. We show that complete synchronization exists as an invariant solution, and give the necessary condition for it to be observed as a stable state. Moreover, in some relevant instances, such a necessary condition takes the form of a Master Stability Function. This generalizes the existing results valid for pairwise interactions to the case of complex systems with the most general possible architecture.

A novel methodology for epidemic risk assessment of COVID-19 outbreak.

Abstract: We propose a novel data-driven framework for assessing the a-priori epidemic risk of a geographical area and for identifying high-risk areas within a country. Our risk index is evaluated as a function of three different components: the hazard of the disease, the exposure of the area and the vulnerability of its inhabitants. As an application, we discuss the case of COVID-19 outbreak in Italy. We characterize each of the twenty Italian regions by using available historical data on air pollution, human mobility, winter temperature, housing concentration, health care density, population size and age. We find that the epidemic risk is higher in some of the Northern regions with respect to Central and Southern Italy. The corresponding risk index shows correlations with the available official data on the number of infected individuals, patients in intensive care and deceased patients, and can help explaining why regions such as Lombardia, Emilia-Romagna, Piemonte and Veneto have suffered much more than the rest of the country. Although the COVID-19 outbreak started in both North (Lombardia) and Central Italy (Lazio) almost at the same time, when the first cases were officially certified at the beginning of 2020, the disease has spread faster and with heavier consequences in regions with higher epidemic risk. Our framework can be extended and tested on other epidemic data, such as those on seasonal flu, and applied to other countries. We also present a policy model connected with our methodology, which might help policy-makers to take informed decisions.

Predicting urban innovation from the US Workforce Mobility Network

Abstract: While great emphasis has been placed on the role of social interactions as a driver of innovation growth, very few empirical studies have explicitly investigated the impact of social network structures on the innovation performance of cities. Past research has mostly explored scaling laws of socio-economic outputs of cities as determined by, for example, the single predictor of population. Here, by drawing on a publicly available dataset of the startup ecosystem, we build the first Workforce Mobility Network among metropolitan areas in the US. We found that node centrality computed on this network accounts for most of the variability observed in cities’ innovation performance and significantly outperforms other predictors such as population size or density, suggesting that policies and initiatives aiming at sustaining innovation processes might benefit from fostering professional networks alongside other economic or systemic incentives. As opposed to previous approaches powered by census data, our model can be updated in real-time upon open databases, opening up new opportunities both for researchers in a variety of disciplines to study urban economies in new ways, and for practitioners to design tools for monitoring such economies in real-time.

Network isolators inhibit failure spreading in complex networks.

Abstract: In our daily lives, we rely on the proper functioning of supply networks, from power grids to water transmission systems. A single failure in these critical infrastructures can lead to a complete collapse through a cascading failure mechanism. Counteracting strategies are thus heavily sought after. In this article, we introduce a general framework to analyse the spreading of failures in complex networks and demostrate that not only decreasing but also increasing the connectivity of the network can be an effective method to contain damages. We rigorously prove the existence of certain subgraphs, called network isolators, that can completely inhibit any failure spreading, and we show how to create such isolators in synthetic and real-world networks. The addition of selected links can thus prevent large scale outages as demonstrated for power transmission grids. A single damage can lead to a complete collapse of supply networks due to a cascading failure mechanism. Kaiser et al. show that by adding new connections network isolators can be created, that can inhibit failure spreading relevant for power grids and water transmission systems.

Influential groups for seeding and sustaining hypergraph contagions

Abstract: Several biological and social contagion phenomena, such as superspreading events or social reinforcement, are the results of multi-body interactions, for which hypergraphs offer a natural mathematical description. In this paper, we develop a novel mathematical framework based on approximate master equations to study contagions on random hypergraphs with a heterogeneous structure, both in terms of group size (hyperedge cardinality) and of membership of nodes to groups (hyperdegree). The characterization of the inner dynamics of groups provides an accurate description of the contagion process, without losing the analytical tractability. Using a contagion model where multi-body interactions are mapped onto a nonlinear infection rate, our two main results show how large groups are influential, in the sense that they drive both the early spread of a contagion and its endemic state (i.e., its stationary state). First, we provide a detailed characterization of the phase transition, which can be continuous or discontinuous with a bistable regime, and derive analytical expressions for the critical and tricritical points. We find that large values of the third moment of the membership distribution suppress the emergence of a discontinuous phase transition. Furthermore, the combination of heterogeneous group sizes and nonlinear contagion facilitates the onset of a mesoscopic localization phase, where contagion is sustained only by the largest groups, thereby inhibiting bistability as well. Second, we formulate a simple problem of optimal seeding for hypergraph contagions to compare two strategies: tuning the allocation of seeds according to either node individual properties or according to group properties. We find that, when the contagion is sufficiently nonlinear, groups are more effective seeds of contagion than individual nodes.

Interdisciplinary researchers attain better performance in funding.

Abstract: Interdisciplinary research is fundamental when it comes to tackling complex problems in our highly interlinked world, and is on the rise globally. Yet, it is unclear why--in an increasingly competitive academic environment--one should pursue an interdisciplinary career given its recent negative press. Several studies have indeed shown that interdisciplinary research often achieves lower impact compared to more specialized work, and is less likely to attract funding. We seek to reconcile such evidence by analyzing a dataset of 44,419 research grants awarded between 2006 and 2018 from the seven national research councils in the UK. We compared the research performance of researchers with an interdisciplinary funding track record with those who have a specialized profile. We found that the former dominates the network of academic collaborations, both in terms of centrality and knowledge brokerage; but such a competitive advantage does not immediately translate into impact. Indeed, by means of a matched pair experimental design, we found that researchers who transcend between disciplines on average achieve lower impacts in their publications than the subject specialists in the short run, but eventually outperform them in funding performance, both in terms of volume and value. Our results suggest that launching an interdisciplinary career may require more time and persistence to overcome extra challenges, but can pave the way for a more successful endeavour.

Individual- and pair-based models of epidemic spreading: master equations and analysis of their forecasting capabilities

Abstract: Mathematical modeling of disease spreading plays a crucial role in understanding, controlling and preventing epidemic outbreaks. In a microscopic description of the propagation of a disease over the complex network of human contacts, the probability that an individual is in a given state (susceptible, infectious, recovered etc) depends on the state of its neighbors in the network. Thus it depends on the state of pairs of nodes, which in turns depends on triples, in a hierarchy of dynamical dependencies. In order to produce models that are at the same time reliable and manageable, one has to understand how to truncate such a hierarchy, and how the chosen order of approximation affects the ability of the model to forecast the real temporal evolution of an epidemics. In this paper we provide a systematic analysis of the reliability (under different hypotheses on the quantity and quality of available data) of the predictions obtained by truncating the hierarchy either at the level of individuals or at the level of pairs. We find that pair-based models are reliable tools both for estimating the epidemiological parameters and for forecasting the temporal evolution of the epidemics, under all conditions taken into account in our work. However, a pair-based approach provides a much better prediction of an individual-based one, only if better data, namely information on the state of node pairs, are available. Overall, our results suggest that more refined mathematical models need to be informed by improved contact tracing techniques to better support decision on policies and containment measures to adopt.

Control technique for synchronization of selected nodes in directed networks

Abstract: In this Letter we propose a method to control a set of arbitrary nodes in a directed network such that they follow a synchronous trajectory which is, in general, not shared by the other units of the network. The problem is inspired to those natural or artificial networks whose proper operating conditions are associated to the presence of clusters of synchronous nodes. Our proposed method is based on the introduction of distributed controllers that modify the topology of the connections in order to generate outer symmetries in the nodes to be controlled. An optimization problem for the selection of the controllers, which includes as a special case the minimization of the number of the links added or removed, is also formulated and an algorithm for its solution is introduced.

Artificial intelligence applied to bailout decisions in financial systemic risk management.

Abstract: We describe the bailout of banks by governments as a Markov Decision Process (MDP) where the actions are equity investments The underlying dynamics is derived from the network of financial institutions linked by mutual exposures, and the negative rewards are associated to the banks' default Each node represents a bank and is associated to a probability of default per unit time (PD) that depends on its capital and is increased by the default of neighbouring nodes Governments can control the systemic risk of the network by providing additional capital to the banks, lowering their PD at the expense of an increased exposure in case of their failure Considering the network of European global systemically important institutions, we find the optimal investment policy that solves the MDP, providing direct indications to governments and regulators on the best way of action to limit the effects of financial crises

On the Dual Nature of Adoption Processes in Complex Networks

Abstract: Adoption processes in socio-technological systems have been widely studied both empirically and theoretically. The way in which social norms, behaviors, and even items such as books, music or other commercial or technological product spread in a population is usually modeled as a process of social contagion in which the agents of a social system can infect their neighbors on the underlying network of social contacts. More recently, various models have also been proposed to reproduce the typical dynamics of a process of discovery, in which an agent explores a space of relations between ideas or items in search for novelties. In both types of processes, the structure of the underlying networks, respectively the network of social contacts in the first case, and the network of relations among items in the second one, plays a fundamental role. However, the two processes have been traditionally seen and studied independently. Here, we provide a brief overview of the existing models of social spreading and exploration and of the latest advancements in both directions. We propose to look at them as two complementary aspects of the same adoption process: on one hand there are items spreading over a social network of individuals influencing each others, on the other hand individuals explore a network of similarities among items to adopt. The two-fold nature of the approach proposed opens up new stimulating challenges for the scientific community of network and data scientists. We conclude by outlining some possible directions that we believe may be relevant to explore in the coming years.

The physics of higher-order interactions in complex systems.

Abstract: Complex networks have become the main paradigm for modelling the dynamics of interacting systems. However, networks are intrinsically limited to describing pairwise interactions, whereas real-world systems are often characterized by higher-order interactions involving groups of three or more units. Higher-order structures, such as hypergraphs and simplicial complexes, are therefore a better tool to map the real organization of many social, biological and man-made systems. Here, we highlight recent evidence of collective behaviours induced by higher-order interactions, and we outline three key challenges for the physics of higher-order systems.

Resilience and fluctuations shape primal and dual communities in spatial networks

Abstract: Both human-made and natural supply networks are built to operate reliably in changing conditions full of external stimuli. Many of these spatial networks exhibit community structures. Here, we show the existence of a second class of communities. These dual communities are based on an exceptionally strong mutual connectivity and can be found for example in leaf venation networks. We demonstrate that traditional and dual communities emerge naturally as two different phases of optimised network structures that are shaped by fluctuations and that they suppress failure spreading, which underlines their importance in understanding the shape of real-world supply networks.

Evolutionary game model of risk propensity in group decision making.

Abstract: We introduce an evolutionary game on hypergraphs in which decisions between a risky alternative and a safe one are taken in social groups of different sizes. The model naturally reproduces choice shifts, namely the differences between the preference of individual decision makers and the consensual choice of a group, that have been empirically observed in choice dilemmas. In particular, a deviation from the Nash equilibrium towards the risky strategy occurs when the dynamics takes place on heterogeneous hypergraphs. These results can explain the emergence of irrational herding and radical behaviours in social groups.

Memory order decomposition of symbolic sequences.

Abstract: We introduce a general method for the study of memory in symbolic sequences based on higher-order Markov analysis. The Markov process that best represents a sequence is expressed as a mixture of matrices of minimal orders, enabling the definition of the so-called memory profile, which unambiguously reflects the true order of correlations. The method is validated by recovering the memory profiles of tunable synthetic sequences. Finally, we scan real data and showcase with practical examples how our protocol can be used to extract relevant stochastic properties of symbolic sequences.

Characterizing Learning Dynamics of Deep Neural Networks via Complex Networks.

Abstract: In this paper, we interpret Deep Neural Networks with Complex Network Theory. Complex Network Theory (CNT) represents Deep Neural Networks (DNNs) as directed weighted graphs to study them as dynamical systems. We efficiently adapt CNT measures to examine the evolution of the learning process of DNNs with different initializations and architectures: we introduce metrics for nodes/neurons and layers, namely Nodes Strength and Layers Fluctuation. Our framework distills trends in the learning dynamics and separates low from high accurate networks. We characterize populations of neural networks (ensemble analysis) and single instances (individual analysis). We tackle standard problems of image recognition, for which we show that specific learning dynamics are indistinguishable when analysed through the solely Link-Weights analysis. Further, Nodes Strength and Layers Fluctuations make unprecedented behaviours emerge: accurate networks, when compared to under-trained models, show substantially divergent distributions with the greater extremity of deviations. On top of this study, we provide an efficient implementation of the CNT metrics for both Convolutional and Fully Connected Networks, to fasten the research in this direction.