-
Gaussian Approximation and Moderate Deviations of Poisson Shot Noises with Application to Compound Generalized Hawkes Processes
Authors:
Mahmoud Khabou,
Giovanni Luca Torrisi
Abstract:
In this article, we give explicit bounds on the Wasserstein and the Kolmogorov distances between random variables lying in the first chaos of the Poisson space and the standard Normal distribution, using the results proved by Last, Peccati and Schulte. Relying on the theory developed in the work of Saulis and Statulevicius and on a fine control of the cumulants of the first chaoses, we also derive…
▽ More
In this article, we give explicit bounds on the Wasserstein and the Kolmogorov distances between random variables lying in the first chaos of the Poisson space and the standard Normal distribution, using the results proved by Last, Peccati and Schulte. Relying on the theory developed in the work of Saulis and Statulevicius and on a fine control of the cumulants of the first chaoses, we also derive moderate deviation principles, Bernstein-type concentration inequalities and Normal approximation bounds with Cramér correction terms for the same variables. The aforementioned results are then applied to Poisson shot-noise processes and, in particular, to the generalized compound Hawkes point processes (a class of stochastic models, introduced in this paper, which generalizes classical Hawkes processes). This extends the recent results availale in the literature regarding the Normal approximation and moderate deviations.
△ Less
Submitted 10 September, 2024;
originally announced September 2024.
-
Competing bootstrap processes on the random graph $G(n,p)$
Authors:
Michele Garetto,
Emilio Leonardi,
Giovanni Luca Torrisi
Abstract:
We consider a generalization of classic bootstrap percolation in which two competing processes concurrently evolve on the same graph $G(n,p)$. Nodes can be in one of three states, conveniently represented by different colors: red, black and white. Initially, a given number $a_R$ of active red nodes (red seeds) are selected uniformly at random among the $n$ nodes. Similarly, a given number $a_B$ of…
▽ More
We consider a generalization of classic bootstrap percolation in which two competing processes concurrently evolve on the same graph $G(n,p)$. Nodes can be in one of three states, conveniently represented by different colors: red, black and white. Initially, a given number $a_R$ of active red nodes (red seeds) are selected uniformly at random among the $n$ nodes. Similarly, a given number $a_B$ of active black nodes (black seeds) are selected uniformly at random among the other $n-a_R$ nodes. All remaining nodes are initially white (inactive). White nodes wake up at times dictated by independent Poisson clocks of rate 1. When a white node wakes up, it checks the state of its neighbors: if the number of red (black) neighbors exceeds the number of black (red) neighbors by a fixed amount $r \geq 2$, the node becomes an active red (black) node, and remains so forever. The parameters of the model are, besides $r$ (fixed) and $n$ (tending to $\infty$), the numbers $a_R$ ($a_B$) of initial red (black) seeds, and the edge existence probability $p=p(n)$. We study the size $A^*_R$ ($A^*_B$) of the final set of active red (black) nodes, identifying different regimes which are analyzed under suitable time-scales, allowing us to obtain detailed (asymptotic) temporal dynamics of the two concurrent activation processes.
△ Less
Submitted 8 May, 2024; v1 submitted 1 May, 2024;
originally announced May 2024.
-
Large and moderate deviations for Gaussian neural networks
Authors:
Claudio Macci,
Barbara Pacchiarotti,
Giovanni Luca Torrisi
Abstract:
We prove large and moderate deviations for the output of Gaussian fully connected neural networks. The main achievements concern deep neural networks (i.e., when the model has more than one hidden layer) and hold for bounded and continuous pre-activation functions. However, for deep neural networks fed by a single input, we have results even if the pre-activation is ReLU. When the network is shall…
▽ More
We prove large and moderate deviations for the output of Gaussian fully connected neural networks. The main achievements concern deep neural networks (i.e., when the model has more than one hidden layer) and hold for bounded and continuous pre-activation functions. However, for deep neural networks fed by a single input, we have results even if the pre-activation is ReLU. When the network is shallow (i.e., there is exactly one hidden layer) the large and moderate principles hold for quite general pre-activation functions.
△ Less
Submitted 11 December, 2024; v1 submitted 3 January, 2024;
originally announced January 2024.
-
Normal approximation of Random Gaussian Neural Networks
Authors:
Nicola Apollonio,
Daniela De Canditiis,
Giovanni Franzina,
Paola Stolfi,
Giovanni Luca Torrisi
Abstract:
In this paper we provide explicit upper bounds on some distances between the (law of the) output of a random Gaussian NN and (the law of) a random Gaussian vector. Our results concern both shallow random Gaussian neural networks with univariate output and fully connected and deep random Gaussian neural networks, with a rather general activation function. The upper bounds show how the widths of the…
▽ More
In this paper we provide explicit upper bounds on some distances between the (law of the) output of a random Gaussian NN and (the law of) a random Gaussian vector. Our results concern both shallow random Gaussian neural networks with univariate output and fully connected and deep random Gaussian neural networks, with a rather general activation function. The upper bounds show how the widths of the layers, the activation functions and other architecture parameters affect the Gaussian approximation of the ouput. Our techniques, relying on Stein's method and integration by parts formulas for the Gaussian law, yield estimates on distances which are indeed integral probability metrics, and include the total variation and the convex distances. These latter metrics are defined by testing against indicator functions of suitable measurable sets, and so allow for accurate estimates of the probability that the output is localized in some region of the space. Such estimates have a significant interest both from a practitioner's and a theorist's perspective.
△ Less
Submitted 20 September, 2023; v1 submitted 10 July, 2023;
originally announced July 2023.
-
Bootstrap percolation on the stochastic block model
Authors:
Giovanni Luca Torrisi,
Michele Garetto,
Emilio Leonardi
Abstract:
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 the…
▽ More
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.
△ Less
Submitted 31 January, 2022;
originally announced January 2022.
-
Asymptotic analysis of Poisson shot noise processes, and applications
Authors:
Giovanni Luca Torrisi,
Emilio Leonardi
Abstract:
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 literatu…
▽ More
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.
△ Less
Submitted 11 August, 2021; v1 submitted 4 August, 2021;
originally announced August 2021.
-
A time-modulated Hawkes process to model the spread of COVID-19 and the impact of countermeasures
Authors:
Michele Garetto,
Emilio Leonardi,
Giovanni Luca Torrisi
Abstract:
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-vary…
▽ More
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.
△ Less
Submitted 2 January, 2021;
originally announced January 2021.
-
Functional inequalities for marked point processes
Authors:
Ian Flint,
Nicolas Privault,
Giovanni Luca Torrisi
Abstract:
In recent years, a number of functional inequalities have been derived for Poisson random measures, with a wide range of applications. In this paper, we prove that such inequalities can be extended to the setting of marked temporal point processes, under mild assumptions on their Papangelou conditional intensity. First, we derive a Poincaré inequality. Second, we prove two transportation cost ineq…
▽ More
In recent years, a number of functional inequalities have been derived for Poisson random measures, with a wide range of applications. In this paper, we prove that such inequalities can be extended to the setting of marked temporal point processes, under mild assumptions on their Papangelou conditional intensity. First, we derive a Poincaré inequality. Second, we prove two transportation cost inequalities. The first one refers to functionals of marked point processes with a Papangelou conditional intensity and is new even in the setting of Poisson random measures. The second one refers to the law of marked temporal point processes with a Papangelou conditional intensity, and extends a related inequality which is known to hold on a general Poisson space. Finally, we provide a variational representation of the Laplace transform of functionals of marked point processes with a Papangelou conditional intensity. The proofs make use of an extension of the Clark-Ocone formula to marked temporal point processes. Our results are shown to apply to classes of renewal, nonlinear Hawkes and Cox point processes.
△ Less
Submitted 14 June, 2019;
originally announced June 2019.
-
Bootstrap percolation on the stochastic block model with k communities
Authors:
Giovanni Luca Torrisi,
Michele Garetto,
Emilio Leonardi
Abstract:
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 communitie…
▽ More
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.
△ Less
Submitted 24 September, 2020; v1 submitted 21 December, 2018;
originally announced December 2018.
-
A large deviation approach to super-critical bootstrap percolation on the random graph $G_{n,p}$
Authors:
Giovanni Luca Torrisi,
Michele Garetto,
Emilio Leonardi
Abstract:
We consider the Erdös--Rényi random graph $G_{n,p}$ and we analyze the simple irreversible epidemic process on the graph, known in the literature as bootstrap percolation. We give a quantitative version of some results by Janson et al. (2012), providing a fine asymptotic analysis of the final size $A_n^*$ of active nodes, under a suitable super-critical regime. More specifically, we establish larg…
▽ More
We consider the Erdös--Rényi random graph $G_{n,p}$ and we analyze the simple irreversible epidemic process on the graph, known in the literature as bootstrap percolation. We give a quantitative version of some results by Janson et al. (2012), providing a fine asymptotic analysis of the final size $A_n^*$ of active nodes, under a suitable super-critical regime. More specifically, we establish large deviation principles for the sequence of random variables $\{\frac{n- A_n^*}{f(n)}\}_{n\geq 1}$ with explicit rate functions and allowing the scaling function $f$ to vary in the widest possible range.
△ Less
Submitted 16 January, 2020; v1 submitted 6 February, 2018;
originally announced February 2018.
-
Gaussian approximation of nonlinear Hawkes processes
Authors:
Giovanni Luca Torrisi
Abstract:
We give a general Gaussian bound for the first chaos (or innovation) of point processes with stochastic intensity constructed by embedding in a bivariate Poisson process. We apply the general result to nonlinear Hawkes processes, providing quantitative central limit theorems.
We give a general Gaussian bound for the first chaos (or innovation) of point processes with stochastic intensity constructed by embedding in a bivariate Poisson process. We apply the general result to nonlinear Hawkes processes, providing quantitative central limit theorems.
△ Less
Submitted 28 September, 2016;
originally announced September 2016.
-
Stochastic dynamics of determinantal processes by integration by parts
Authors:
Laurent Decreusefond,
Ian Flint,
Nicolas Privault,
Giovanni Luca Torrisi
Abstract:
We derive an integration by parts formula for functionals of determinantal processes on compact sets, completing the arguments of [4]. This is used to show the existence of a configuration-valued diffusion process which is non-colliding and admits the distribution of the determinantal process as reversible law. In particular, this approach allows us to build a concrete example of the associated di…
▽ More
We derive an integration by parts formula for functionals of determinantal processes on compact sets, completing the arguments of [4]. This is used to show the existence of a configuration-valued diffusion process which is non-colliding and admits the distribution of the determinantal process as reversible law. In particular, this approach allows us to build a concrete example of the associated diffusion process, providing an illustration of the results of [4] and [30].
△ Less
Submitted 29 September, 2015; v1 submitted 22 October, 2012;
originally announced October 2012.
-
Load optimization in a planar network
Authors:
Charles Bordenave,
Giovanni Luca Torrisi
Abstract:
We analyze the asymptotic properties of a Euclidean optimization problem on the plane. Specifically, we consider a network with three bins and $n$ objects spatially uniformly distributed, each object being allocated to a bin at a cost depending on its position. Two allocations are considered: the allocation minimizing the bin loads and the allocation allocating each object to its less costly bin.…
▽ More
We analyze the asymptotic properties of a Euclidean optimization problem on the plane. Specifically, we consider a network with three bins and $n$ objects spatially uniformly distributed, each object being allocated to a bin at a cost depending on its position. Two allocations are considered: the allocation minimizing the bin loads and the allocation allocating each object to its less costly bin. We analyze the asymptotic properties of these allocations as the number of objects grows to infinity. Using the symmetries of the problem, we derive a law of large numbers, a central limit theorem and a large deviation principle for both loads with explicit expressions. In particular, we prove that the two allocations satisfy the same law of large numbers, but they do not have the same asymptotic fluctuations and rate functions.
△ Less
Submitted 21 December, 2010; v1 submitted 25 February, 2009;
originally announced February 2009.
-
Large deviations of Poisson cluster processes
Authors:
Charles Bordenave,
Giovanni Luca Torrisi
Abstract:
In this paper we prove scalar and sample path large deviation principles for a large class of Poisson cluster processes. As a consequence, we provide a large deviation principle for ergodic Hawkes point processes.
In this paper we prove scalar and sample path large deviation principles for a large class of Poisson cluster processes. As a consequence, we provide a large deviation principle for ergodic Hawkes point processes.
△ Less
Submitted 11 April, 2007;
originally announced April 2007.
