arXiv subjects

Emilio Leonardi

Publications and source records attributed to Emilio Leonardi.

At least 19 recordsLinked to original sources

arXiv2025

Dominance or Fair Play in Social Networks? A Model of Influencer Popularity Dynamic

This paper presents a data-driven mean-field approach to model the popularity dynamics of users seeking public attention, i.e., influencers. We propose a novel analytical model that integrates individual activity patterns, expertise in producing viral content, exogenous events, and the platform's role in visibility enhancement, ultimately determining each influencer's success. We analytically derive sufficient conditions for system ergodicity, enabling predictions of popularity distributions. A sensitivity analysis explores various system configurations, highlighting conditions favoring either dominance or fair play among influencers. Our findings offer valuable insights into the potential evolution of social networks towards more equitable or biased influence ecosystems.

cs.SI↗

arXiv2025

Information Retrieval in the Age of Generative AI: The RGB Model

The advent of Large Language Models (LLMs) and generative AI is fundamentally transforming information retrieval and processing on the Internet, bringing both great potential and significant concerns regarding content authenticity and reliability. This paper presents a novel quantitative approach to shed light on the complex information dynamics arising from the growing use of generative AI tools. Despite their significant impact on the digital ecosystem, these dynamics remain largely uncharted and poorly understood. We propose a stochastic model to characterize the generation, indexing, and dissemination of information in response to new topics. This scenario particularly challenges current LLMs, which often rely on real-time Retrieval-Augmented Generation (RAG) techniques to overcome their static knowledge limitations. Our findings suggest that the rapid pace of generative AI adoption, combined with increasing user reliance, can outpace human verification, escalating the risk of inaccurate information proliferation across digital resources. An in-depth analysis of Stack Exchange data confirms that high-quality answers inevitably require substantial time and human effort to emerge. This underscores the considerable risks associated with generating persuasive text in response to new questions and highlights the critical need for responsible development and deployment of future generative AI tools.

cs.IR↗

arXiv2025

Scalable Decentralized Algorithms for Online Personalized Mean Estimation

In numerous settings, agents lack sufficient data to directly learn a model. Collaborating with other agents may help, but it introduces a bias-variance trade-off, when local data distributions differ. A key challenge is for each agent to identify clients with similar distributions while learning the model, a problem that remains largely unresolved. This study focuses on a simplified version of the overarching problem, where each agent collects samples from a real-valued distribution over time to estimate its mean. Existing algorithms face impractical space and time complexities (quadratic in the number of agents A). To address scalability challenges, we propose a framework where agents self-organize into a graph, allowing each agent to communicate with only a selected number of peers r. We introduce two collaborative mean estimation algorithms: one draws inspiration from belief propagation, while the other employs a consensus-based approach, with complexity of O( r |A| log |A|) and O(r |A|), respectively. We establish conditions under which both algorithms yield asymptotically optimal estimates and offer a theoretical characterization of their performance.

cs.LG↗

arXiv2024

Competing bootstrap processes on the random graph $G(n,p)$

We introduce and analyze a competitive extension of classical bootstrap percolation on the Erd\H{o}s--R\'{e}nyi random graph $G(n,p_n)$. Nodes can be red, black, or white, with red and black representing two competing active states. Starting from two sets of initially active seeds, white nodes activate according to independent Poisson clocks and become permanently red or black whenever the number of active neighbors of one color exceeds that of the competing color by at least a fixed threshold $r\geq2$. We characterize the asymptotic dynamics and final sizes of the two competing activation processes over all relevant seed-density scales. Our results reveal a sharp qualitative dichotomy. In the sub-critical regime, competition is asymptotically negligible at first order: each process reaches the same normalized final size as it would in the absence of the competing process. In the super-critical regime, instead, the initial advantage of the process with the larger seed density is amplified: the dominant process activates $n-o(n)$ nodes, while the competing process is suppressed and remains confined to a much smaller scale. Beyond final-size asymptotics, we derive a fluid-limit description of the joint activation trajectories and characterize the relevant activation time scales. The analysis combines uniform concentration estimates, deterministic limiting Cauchy problems, stochastic coupling, and multiscale arguments.

math.PR↗

arXiv2023

Ranking a Set of Objects using Heterogeneous Workers: QUITE an Easy Problem

We focus on the problem of ranking $N$ objects starting from a set of noisy pairwise comparisons provided by a crowd of unequal workers, each worker being characterized by a specific degree of reliability, which reflects her ability to rank pairs of objects. More specifically, we assume that objects are endowed with intrinsic qualities and that the probability with which an object is preferred to another depends both on the difference between the qualities of the two competitors and on the reliability of the worker. We propose QUITE, a non-adaptive ranking algorithm that jointly estimates workers' reliabilities and qualities of objects. Performance of QUITE is compared in different scenarios against previously proposed algorithms. Finally, we show how QUITE can be naturally made adaptive.

cs.LG↗

arXiv2023

Planning interventions in a controlled pandemic: the COVID-19 case

Restrictions on social and economic activities, as well as vaccinations, have been a key intervention in containing the COVID-19 epidemic. Our work focuses on better understanding the options available to policymakers under the conditions and uncertainties created by the onset of a new pandemic. More precisely, we focus on two control strategies. The first aims to control the rate of new infections to prevent congestion of the health care system. The latter directly controls hospitalizations and intensive care units (ICUs) occupation. By a first-order analysis, we show that, on the one hand, due to the difficulty in contact tracing and the lack of accurate information, controlling the transmission rate may be difficult, leading to instability. On the other hand, although hospitalizations and ICUs are easily accessible and less noisy than the rate of new infections, a delay is introduced in the control loop, which may endanger system stability. Our framework allows assessing the impact on economic and social costs of the above strategies in a scenario enriched by: i) population heterogeneity in terms of mortality rate and risk exposure, ii) closed-loop control of the epidemiological curve, and iii) progressive vaccination of individuals.

math.OC↗

arXiv2023

Modeling communication asymmetry and content personalization in online social networks

The increasing popularity of online social networks (OSNs) attracted growing interest in modeling social interactions. On online social platforms, a few individuals, commonly referred to as influencers, produce the majority of content consumed by users and hegemonize the landscape of the social debate. However, classical opinion models do not capture this communication asymmetry. We develop an opinion model inspired by observations on social media platforms {with two main objectives: first, to describe this inherent communication asymmetry in OSNs, and second, to model the effects of content personalization. We derive a Fokker-Planck equation for the temporal evolution of users' opinion distribution and analytically characterize the stationary system behavior. Analytical results, confirmed by Monte-Carlo simulations, show how strict forms of content personalization tend to radicalize user opinion, leading to the emergence of echo chambers, and favor structurally advantaged influencers. As an example application, we apply our model to Facebook data during the Italian government crisis in the summer of 2019. Our work provides a flexible framework to evaluate the impact of content personalization on the opinion formation process, focusing on the interaction between influential individuals and regular users. This framework is interesting in the context of marketing and advertising, misinformation spreading, politics and activism.

cs.SI↗

arXiv2023

On the Robustness of Topics API to a Re-Identification Attack

Web tracking through third-party cookies is considered a threat to users' privacy and is supposed to be abandoned in the near future. Recently, Google proposed the Topics API framework as a privacy-friendly alternative for behavioural advertising. Using this approach, the browser builds a user profile based on navigation history, which advertisers can access. The Topics API has the possibility of becoming the new standard for behavioural advertising, thus it is necessary to fully understand its operation and find possible limitations. This paper evaluates the robustness of the Topics API to a re-identification attack where an attacker reconstructs the user profile by accumulating user's exposed topics over time to later re-identify the same user on a different website. Using real traffic traces and realistic population models, we find that the Topics API mitigates but cannot prevent re-identification to take place, as there is a sizeable chance that a user's profile is unique within a website's audience. Consequently, the probability of correct re-identification can reach 15-17%, considering a pool of 1,000 users. We offer the code and data we use in this work to stimulate further studies and the tuning of the Topic API parameters.

cs.CY↗

arXiv2023

Federated Learning under Heterogeneous and Correlated Client Availability

The enormous amount of data produced by mobile and IoT devices has motivated the development of federated learning (FL), a framework allowing such devices (or clients) to collaboratively train machine learning models without sharing their local data. FL algorithms (like FedAvg) iteratively aggregate model updates computed by clients on their own datasets. Clients may exhibit different levels of participation, often correlated over time and with other clients. This paper presents the first convergence analysis for a FedAvg-like FL algorithm under heterogeneous and correlated client availability. Our analysis highlights how correlation adversely affects the algorithm's convergence rate and how the aggregation strategy can alleviate this effect at the cost of steering training toward a biased model. Guided by the theoretical analysis, we propose CA-Fed, a new FL algorithm that tries to balance the conflicting goals of maximizing convergence speed and minimizing model bias. To this purpose, CA-Fed dynamically adapts the weight given to each client and may ignore clients with low availability and large correlation. Our experimental results show that CA-Fed achieves higher time-average accuracy and a lower standard deviation than state-of-the-art AdaFed and F3AST, both on synthetic and real datasets.

cs.LG↗

arXiv2022

Bootstrap percolation on the stochastic block model

We analyze the bootstrap percolation process on the stochastic block model (SBM), a natural extension of the Erdős--Rényi random graph that incorporates the community structure observed in many real systems. In the SBM, nodes are partitioned into two subsets, which represent different communities, and pairs of nodes are independently connected with a probability that depends on the communities they belong to. Under mild assumptions on the system parameters, we prove the existence of a sharp phase transition for the final number of active nodes and characterize the sub-critical and the super-critical regimes in terms of the number of initially active nodes, which are selected uniformly at random in each community.

math.PR↗

arXiv2021

Asymptotic analysis of Poisson shot noise processes, and applications

Poisson shot noise processes are natural generalizations of compound Poisson processes that have been widely applied in insurance, neuroscience, seismology, computer science and epidemiology. In this paper we study sharp deviations, fluctuations and the stable probability approximation of Poisson shot noise processes. Our achievements extend, improve and complement existing results in the literature. We apply the theoretical results to Poisson cluster point processes, including generalized linear Hawkes processes, and risk processes with delayed claims. Many examples are discussed in detail.

math.PR↗

arXiv2021

Similarity Caching: Theory and Algorithms

This paper focuses on similarity caching systems, in which a user request for an {object~$o$} that is not in the cache can be (partially) satisfied by a similar stored {object~$o'$}, at the cost of a loss of user utility. Similarity caching systems can be effectively employed in several application areas, like multimedia retrieval, recommender systems, genome study, and machine learning training/serving. However, despite their relevance, the behavior of such systems is far from being well understood. In this paper, we provide a first comprehensive analysis of similarity caching in the offline, adversarial, and stochastic settings. We show that similarity caching raises significant new challenges, for which we propose the first dynamic policies with some optimality guarantees. We evaluate the performance of our schemes under both synthetic and real request traces.

cs.NI↗

arXiv2021

A Swiss Army Knife for Dynamic Caching in Small Cell Networks

We consider a dense cellular network, in which a limited-size cache is available at every base station (BS). Coordinating content allocation across the different caches can lead to significant performance gains, but is a difficult problem even when full information about the network and the request process is available. In this paper we present qLRU-$Δ$, a general-purpose dynamic caching policy that can be tailored to optimize different performance metrics also in presence of coordinated multipoint transmission techniques. The policy requires neither direct communication among BSs, nor a priori knowledge of content popularity and, under stationary request processes, has provable performance guarantees.

cs.NI↗

arXiv2021

Content Placement in Networks of Similarity Caches

Similarity caching systems have recently attracted the attention of the scientific community, as they can be profitably used in many application contexts, like multimedia retrieval, advertising, object recognition, recommender systems and online content-match applications. In such systems, a user request for an object $o$, which is not in the cache, can be (partially) satisfied by a similar stored object $o$', at the cost of a loss of user utility. In this paper we make a first step into the novel area of similarity caching networks, where requests can be forwarded along a path of caches to get the best efficiency-accuracy tradeoff. The offline problem of content placement can be easily shown to be NP-hard, while different polynomial algorithms can be devised to approach the optimal solution in discrete cases. As the content space grows large, we propose a continuous problem formulation whose solution exhibits a simple structure in a class of tree topologies. We verify our findings using synthetic and realistic request traces.

cs.NI↗

arXiv2021

Asynchronous semi-anonymous dynamics over large-scale networks

We analyze a class of stochastic processes, referred to as asynchronous and semi-anonymous dynamics (ASD), over directed labeled random networks. These processes are a natural tool to describe general best-response and noisy best-response dynamics in network games where each agent, at random times governed by independent Poisson clocks, can choose among a finite set of actions. The payoff is determined by the relative popularity of different actions among neighbors, while being independent of the specific identities of neighbors. Using a mean-field approach, we prove that, under certain conditions on the network and initial node configuration, the evolution of ASD can be approximated, in the limit of large network sizes, by the solution of a system of non-linear ordinary differential equations. Our framework is very general and applies to a large class of graph ensembles for which the typical random graph locally behaves like a tree. In particular, we will focus on labeled configuration-model random graphs, a generalization of the traditional configuration model which allows different classes of nodes to be mixed together in the network, permitting us, for example, to incorporate a community structure in the system. Our analysis also applies to configuration-model graphs having a power-law degree distribution, an essential feature of many real systems. To demonstrate the power and flexibility of our framework, we consider several examples of dynamics belonging to our class of stochastic processes. Moreover, we illustrate by simulation the applicability of our analysis to realistic scenarios by running our example dynamics over a real social network graph.

cs.SI↗

arXiv2021

A time-modulated Hawkes process to model the spread of COVID-19 and the impact of countermeasures

Motivated by the recent outbreak of coronavirus (COVID-19), we propose a stochastic model of epidemic temporal growth and mitigation based on a time-modulated Hawkes process. The model is sufficiently rich to incorporate specific characteristics of the novel coronavirus, to capture the impact of undetected, asymptomatic and super-diffusive individuals, and especially to take into account time-varying counter-measures and detection efforts. Yet, it is simple enough to allow scalable and efficient computation of the temporal evolution of the epidemic, and exploration of what-if scenarios. Compared to traditional compartmental models, our approach allows a more faithful description of virus specific features, such as distributions for the time spent in stages, which is crucial when the time-scale of control (e.g., mobility restrictions) is comparable to the lifetime of a single infection. We apply the model to the first and second wave of COVID-19 in Italy, shedding light into several effects related to mobility restrictions introduced by the government, and to the effectiveness of contact tracing and mass testing performed by the national health service.

q-bio.PE↗

arXiv2020

Bootstrap percolation on the stochastic block model with k communities

We analyze the bootstrap percolation process on the stochastic block model (SBM), a natural extension of the Erdös--Rényi random graph that allows representing the "community structure" observed in many real systems. In the SBM, nodes are partitioned into subsets, which represent different communities, and pairs of nodes are independently connected with a probability that depends on the communities they belong to. Under mild assumptions on system parameters, we prove the existence of a sharp phase transition for the final number of active nodes and characterize sub-critical and super-critical regimes in terms of the number of initially active nodes, which are selected uniformly at random in each community.

math.PR↗

arXiv2020

Ranking a set of objects: a graph based least-square approach

We consider the problem of ranking $N$ objects starting from a set of noisy pairwise comparisons provided by a crowd of equal workers. We assume that objects are endowed with intrinsic qualities and that the probability with which an object is preferred to another depends only on the difference between the qualities of the two competitors. We propose a class of non-adaptive ranking algorithms that rely on a least-squares optimization criterion for the estimation of qualities. Such algorithms are shown to be asymptotically optimal (i.e., they require $O(\frac{N}{ε^2}\log \frac{N}δ)$ comparisons to be $(ε, δ)$-PAC). Numerical results show that our schemes are very efficient also in many non-asymptotic scenarios exhibiting a performance similar to the maximum-likelihood algorithm. Moreover, we show how they can be extended to adaptive schemes and test them on real-world datasets.

cs.IR↗