Showing papers by "Vito Latora published in 2022"

Synchronization induced by directed higher-order interactions

Abstract: Abstract Non-reciprocal interactions play a crucial role in many social and biological complex systems. While directionality has been thoroughly accounted for in networks with pairwise interactions, its effects in systems with higher-order interactions have not yet been explored as deserved. Here, we introduce the concept of M -directed hypergraphs, a general class of directed higher-order structures, which allows to investigate dynamical systems coupled through directed group interactions. As an application we study the synchronization of nonlinear oscillators on 1-directed hypergraphs, finding that directed higher-order interactions can destroy synchronization, but also stabilize otherwise unstable synchronized states.

Turing patterns in systems with high-order interactions

Abstract: Turing theory of pattern formation is among the most popular theoretical means to account for the variety of spatio-temporal structures observed in Nature and, for this reason, finds applications in many different fields. While Turing patterns have been thoroughly investigated on continuous support and on networks, only a few attempts have been made towards their characterization in systems with higher-order interactions. In this paper, we propose a way to include group interactions in reaction-diffusion systems, and we study their effects on the formation of Turing patterns. To achieve this goal, we rewrite the problem originally studied by Turing in a general form that accounts for a microscropic description of interactions of any order in the form of a hypergraph, and we prove that the interplay between the different orders of interaction may either enhance or repress the emergence of Turing patterns. Our results shed light on the mechanisms of pattern-formation in systems with many-body interactions and pave the way for further extensions of Turing original framework.

The shape of memory in temporal networks

Abstract: How to best define, detect and characterize network memory, i.e. the dependence of a network's structure on its past, is currently a matter of debate. Here we show that the memory of a temporal network is inherently multidimensional, and we introduce a mathematical framework for defining and efficiently estimating the microscopic shape of memory, which characterises how the activity of each link intertwines with the activities of all other links. We validate our methodology on a range of synthetic models, and we then study the memory shape of real-world temporal networks spanning social, technological and biological systems, finding that these networks display heterogeneous memory shapes. In particular, online and offline social networks are markedly different, with the latter showing richer memory and memory scales. Our theory also elucidates the phenomenon of emergent virtual loops and provides a novel methodology for exploring the dynamically rich structure of complex systems.

Social synchronization of brain activity increases during eye-contact

Abstract: Abstract Humans make eye-contact to extract information about other people’s mental states, recruiting dedicated brain networks that process information about the self and others. Recent studies show that eye-contact increases the synchronization between two brains but do not consider its effects on activity within single brains. Here we investigate how eye-contact affects the frequency and direction of the synchronization within and between two brains and the corresponding network characteristics. We also evaluate the functional relevance of eye-contact networks by comparing inter- and intra-brain networks of friends vs. strangers and the direction of synchronization between leaders and followers. We show that eye-contact increases higher inter- and intra-brain synchronization in the gamma frequency band. Network analysis reveals that some brain areas serve as hubs linking within- and between-brain networks. During eye-contact, friends show higher inter-brain synchronization than strangers. Dyads with clear leader/follower roles demonstrate higher synchronization from leader to follower in the alpha frequency band. Importantly, eye-contact affects synchronization between brains more than within brains, demonstrating that eye-contact is an inherently social signal. Future work should elucidate the causal mechanisms behind eye-contact induced synchronization.

Social interactions affect discovery processes

Abstract: Our network of acquaintances determines how we get exposed to ideas, products, or cultural artworks (books, music, movies, etc.). Though this principle is part of our common sense, little is known about the specific pathways through which our peers influence our discovery processes and our experience of the new. Here, we fill this gap by investigating a data set containing the whole listening histories of a large, socially connected sample of users from the online music platform \emph{Last.fm}. We demonstrate that users exhibit highly heterogeneous discovery rates of new songs and artists and that their social neighborhood significantly influences their behavior. More explorative users tend to interact with peers more prone to explore new content. We capture this phenomenology in a modeling scheme where users are represented by random walkers exploring a graph of songs or artists and interacting with each other through their social links. Even starting from a uniform population of agents (no natural differences among the individuals), our model predicts the emergence of strong heterogeneous exploration patterns, with users clustered according to their musical tastes and propensity to explore. We contend our approach can pave the way to a quantitative approach to collective discovery processes.

Network analysis of synovial RNA sequencing identifies gene-gene interactions predictive of response in rheumatoid arthritis

Abstract: Abstract Background To determine whether gene-gene interaction network analysis of RNA sequencing (RNA-Seq) of synovial biopsies in early rheumatoid arthritis (RA) can inform our understanding of RA pathogenesis and yield improved treatment response prediction models. Methods We utilized four well curated pathway repositories obtaining 10,537 experimentally evaluated gene-gene interactions. We extracted specific gene-gene interaction networks in synovial RNA-Seq to characterize histologically defined pathotypes in early RA and leverage these synovial specific gene-gene networks to predict response to methotrexate-based disease-modifying anti-rheumatic drug (DMARD) therapy in the Pathobiology of Early Arthritis Cohort (PEAC). Differential interactions identified within each network were statistically evaluated through robust linear regression models. Ability to predict response to DMARD treatment was evaluated by receiver operating characteristic (ROC) curve analysis. Results Analysis comparing different histological pathotypes showed a coherent molecular signature matching the histological changes and highlighting novel pathotype-specific gene interactions and mechanisms. Analysis of responders vs non-responders revealed higher expression of apoptosis regulating gene-gene interactions in patients with good response to conventional synthetic DMARD. Detailed analysis of interactions between pairs of network-linked genes identified the SOCS2/STAT2 ratio as predictive of treatment success, improving ROC area under curve (AUC) from 0.62 to 0.78. We identified a key role for angiogenesis, observing significant statistical interactions between NOS3 (eNOS) and both CAMK1 and eNOS activator AKT3 when comparing responders and non-responders. The ratio of CAMKD2/NOS3 enhanced a prediction model of response improving ROC AUC from 0.63 to 0.73. Conclusions We demonstrate a novel, powerful method which harnesses gene interaction networks for leveraging biologically relevant gene-gene interactions leading to improved models for predicting treatment response.

Maximal dispersion of adaptive random walks

Abstract: ,

The making of the oral microbiome in Agta hunter-gatherers

Abstract: Ecological and genetic factors have influenced the composition of the human microbiome during our evolutionary history. We analyzed the oral microbiota of the Agta, a hunter-gatherer population where part of its members is adopting an agricultural diet. We show that age is the strongest factor modulating the microbiome, likely through immunosenescence as there is an increase of pathogenicity with age. Biological and cultural processes generate sexual dimorphism in the oral microbiome. A small subset of oral bacteria is influenced by the host genome, linking host collagen genes to bacterial biofilm formation. Our data also suggests that shifting from a fish/meat to a rice-rich diet transforms their microbiome, mirroring the Neolithic transition. All these factors have implications in the epidemiology of oral diseases. Thus, the human oral microbiome is multifactorial, and shaped by various ecological and social factors that modify the oral environment.

Hunter-gatherer oral microbiomes are shaped by contact network structure

Abstract: Ancestral humans evolved a complex social structure still observed in extant hunter-gatherers. Here we investigate the effects of extensive sociality and mobility on the oral microbiome of 138 Agta hunter-gatherers from the Philippines. Comparisons of microbiome composition showed that the Agta are more similar to Central African Bayaka hunter-gatherers than to neighboring farmers. We also defined the Agta social microbiome as a set of 137 oral bacteria (only 7% of 1980 amplicon sequence variants) significantly influenced by social contact (quantified through wireless sensors of short-range interactions). We show that interaction networks covering large areas, and their strong links between close kin, spouses, and even unrelated friends, can significantly predict bacterial transmission networks across Agta camps. Finally, more central individuals to social networks are also bacterial supersharers. We conclude that hunter-gatherer social microbiomes, which are predominantly pathogenic, were shaped by evolutionary tradeoffs between extensive sociality and disease spread.

Collective Games on Hypergraphs

Abstract: Human activities often require simultaneous decision-making of individuals in groups. These processes cannot be coherently addressed by means of networks, as networks only allow for pairwise interactions. Here, we propose a general implementation for collective games in which higher-order interactions are encoded on hypergraphs. We employ it for the study of the public goods game by first validating the analytical expression of the replicator dynamics in uniform and heterogeneous populations, and then by introducing a procedure for retrieving empirical synergistic effects of group interactions from real datasets. U. Alvarez-Rodriguez (B) Data Analytics Group, University of Zurich, Zurich, Switzerland F. Battiston Department of Network and Data Science, Central European University, Vienna, Austria Department of Anthropology, University of Zurich, Zurich, Switzerland G. Ferraz de Arruda · Y. Moreno ISI Foundation, Turin, Italy Y. Moreno Institute for Biocomputation and Physics of Complex Systems, University of Zaragoza, Zaragoza 50008, Spain Department of Theoretical Physics, University of Zaragoza, Zaragoza 50009, Spain M. Perc Faculty of Natural Sciences and Mathematics, University of Maribor, Koroška cesta 160, 2000 Maribor, Slovenia Department of Medical Research, China Medical University Hospital, China Medical University, Taichung, Taiwan Complexity Science Hub Vienna, Josefstädterstraße 39, 1080 Vienna, Austria V. Latora School of Mathematical Sciences, Queen Mary University of London, London E1 4NS, UK Dipartimento di Fisica ed Astronomia, Università di Catania and INFN, 95123 Catania, Italy The Alan Turing Institute, The British Library, London NW1 2DB, UK © The Author(s), under exclusive license to Springer Nature Switzerland AG 2022 F. Battiston and G. Petri (eds.), Higher-Order Systems, Understanding Complex Systems, https://doi.org/10.1007/978-3-030-91374-8_15 377 378 U. Alvarez-Rodriguez et al.

An AI approach for managing financial systemic risk via bank bailouts by taxpayers

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.

Deep Neural Networks as Complex Networks

Abstract: Deep Neural Networks are, from a physical perspective, graphs whose ‘links‘ and ‘vertices‘ iteratively process data and solve tasks sub-optimally. We use Complex Network Theory (CNT) to represents Deep Neural Networks (DNNs) as directed weighted graphs: within this framework, we introduce metrics to study DNNs as dynamical systems, with a granularity that spans from weights to layers, including neurons. CNT discriminates networks that differ in the number of parameters and neurons, the type of hidden layers and activations, and the objective task. We further show that our metrics discriminate low vs. high performing networks. CNT is a comprehensive method to reason about DNNs and a complementary approach to explain a model’s behavior that is physically grounded to networks theory and goes beyond the well-studied input-output relation.

Reconstruction of cascading failures in dynamical models of power grids

Abstract: In this article, we propose a method to reconstruct the active links of a power network described by a second-order Kuramoto model and subject to dynamically induced cascading failures. Starting from the assumption (realistic for power grids) that the structure of the network is known, our method reconstructs the active links from the evolution of the relevant dynamical quantities of the nodes of the system, that is, the node phases and angular velocities. We find that, to reconstruct the temporal sequence of the faults, it is crucial to use time series with a small number of samples, as the observation window should be smaller than the temporal distance between subsequent events. This requirement is in contrast with the need of using larger sets of data in the presence of noise, such that the number of samples to feed in the algorithm has to be selected as a trade-off between the prediction error and temporal resolution of the active link reconstruction.

MultiSAGE: a multiplex embedding algorithm for inter-layer link prediction

Abstract: Research on graph representation learning has received great attention in recent years. However, most of the studies so far have focused on the embedding of single-layer graphs. The few studies dealing with the problem of representation learning of multilayer structures rely on the strong hypothesis that the inter-layer links are known, and this limits the range of possible applications. Here we propose MultiSAGE, a generalization of the GraphSAGE algorithm that allows to embed multiplex networks. We show that MultiSAGE is capable to reconstruct both the intra-layer and the inter-layer connectivity, outperforming GraphSAGE, which has been designed for simple graphs. Next, through a comprehensive experimental analysis, we shed light also on the performance of the embedding, both in simple and in multiplex networks, showing that either the density of the graph or the randomness of the links strongly influences the quality of the embedding.

Socially-enhanced discovery processes

Abstract: Gabriele Di Bona, Iacopo Iacopini, Enrico Ubaldi, Vito Latora, and Vittorio Loreto School of Mathematical Science, Queen Mary University of London, London E1 4NS, United Kingdom Department of Network and Data Science, Central European University, 1100 Vienna, Austria SONY Computer Science Laboratories, Paris, 6, rue Amyot, 75005, Paris, France Dipartimento di Fisica ed Astronomia, Università di Catania and INFN, I-95123 Catania, Italy Complexity Science Hub, Josefst äadter Strasse 39, A 1080 Vienna, Austria Sapienza Univ. of Rome, Physics Dept., Piazzale Aldo Moro 2, 00185 Rome, Italy

Model of spatial competition on discrete markets

Abstract: We propose a dynamical model of price formation on a spatial market where sellers and buyers are placed on the nodes of a graph, and the distribution of the buyers depends on the positions and prices of the sellers. We find that, depending on the positions of the sellers and on the level of information available, the price dynamics of our model can either converge to fixed prices, or produce cycles of different amplitudes and periods. We show how to measure the strength of competition in a spatial network by extracting the exponent of the scaling of the prices with the size of the system. As an application, we characterize the different level of competition in street networks of real cities across the globe. Finally, using the model dynamics we can define a novel measure of node centrality, which quantifies the relevance of a node in a competitive market.

Hyper-diffusion on multiplex networks

Abstract: Multiplex networks describe systems whose interactions can be of different nature, and are fundamental to understand complexity of networks beyond the framework of simple graphs. Recently it has been pointed out that restricting the attention to pairwise interactions is also a limitation, as the vast majority of complex systems include higher-order interactions that strongly affect their dynamics. Here, we propose hyper-diffusion on multiplex networks, a dynamical process in which diffusion on each single layer is coupled with the diffusion in other layers thanks to the presence of higher-order interactions occurring when there exists link overlap. We show that hyper-diffusion on a duplex network (a multiplex network with two layers) can be described by the hyper-Laplacian in which the strength of four-body interactions among every set of four replica nodes connected in both layers can be tuned by a parameter δ 11 ⩾ 0. The hyper-Laplacian reduces to the standard lower Laplacian, capturing pairwise interactions at the two layers, when δ 11 = 0. By combining tools of spectral graph theory, applied topology and network science we provide a general understanding of hyper-diffusion on duplex networks when δ 11 > 0, including theoretical bounds on the Fiedler and the largest eigenvalue of hyper-Laplacians and the asymptotic expansion of their spectrum for δ 11 ≪ 1 and δ 11 ≫ 1. Although hyper-diffusion on multiplex networks does not imply a direct ‘transfer of mass’ among the layers (i.e. the average state of replica nodes in each layer is a conserved quantity of the dynamics), we find that the dynamics of the two layers is coupled as the relaxation to the steady state becomes synchronous when higher-order interactions are taken into account and the Fiedler eigenvalue of the hyper-Laplacian is not localized in a single layer of the duplex network.

The decentralized evolution of decentralization across fields: from Governance to Blockchain

Abstract: Decentralization is a widespread concept across disciplines such as Economics, Political Science and Computer Science, which use it in distinct but overlapping ways. Here, we investigate the scholarly history of the term by analysing 425 k academic publications mentioning (de)centralization. We reveal that the fraction of papers on the topic has been exponentially increasing since the 1950s, with 1 author in 154 contributing to a paper on (de)centralization in 2021. We then cluster papers using both semantic and citation information and show that the topic has independently emerged in different fields, while cross-disciplinary contamination started only more recently. Finally, focusing on the two most prominent clusters by number of papers and influence, we show how Blockchain has become the most influential field about 10 years ago, while Governance dominated before the 1990s, and we characterize their interactions with other fields. Our results add a quantitative dimension to the history of a key yet elusive concept. Furthermore, the introduced framework is general and our publicly released pipeline may be used to run similar analyses on other concepts in the academic literature. keywords and the hierarchical tree resulting from the clustering algorithm.