arXiv subjects

Frédéric Giroire

Publications and source records attributed to Frédéric Giroire.

13 recordsLinked to original sources

arXiv2026

Degree-Mass Message Passing for Betweenness Ranking in Directed and Undirected Networks

Computing the importance of nodes in networks is a long-standing fundamental problem that has driven extensive study of various centrality measures. A particularly well-known centrality measure is betweenness centrality, whose exact computation becomes prohibitive on large-scale networks. Graph Neural Network (GNN) models have thus been proposed to predict the ranking of nodes by betweenness centrality. However, existing GNN-based methods either have graph-size-dependent parameter counts or are limited to undirected graphs. We propose a lightweight GNN architecture that exploits the empirically observed relationship between betweenness centrality and multi-hop degree mass. This motivates the use of degree masses as size-invariant node features. To improve generalization, we train on synthetic graphs whose degree distributions more closely match those of real-world networks, including directed and undirected scale-free graphs and uniformly directed hyperbolic random graphs. We evaluate our model on 14 real-world networks spanning eight domains, including social, email, and citation networks, across both directed and undirected regimes. The experiments show that our model improves the Kendall $τ_b$ correlation by up to 24.6\% on undirected and 10.9\% on directed graphs, while using 56$\times$ fewer parameters than the lightest competing GNN baseline and achieving competitive inference time, with up to a 24.5$\times$ speedup on selected directed graphs.

cs.LG↗

arXiv2026

WattLayer: Get Layers Right to Estimate Inference Energy of Neural Networks

The widespread adoption of Artificial Intelligence (AI) has led to increasing concerns about energy consumption, yet there is a lack of standardized methodologies to accurately estimate AI inference energy consumption, particularly across various tasks and architectures. In this study, we propose a task independent, layer-wise energy estimation model for AI architectures. Our model is evaluated on a large dataset of more than 100,000 layers for 295 neural network architectures across 3 widely-used tasks and 3 distinct hardware platforms. Our approach achieves a median error of 19.6%, outperforming state-of-the-art methods. We further show that layer-wise decomposition generalize to new tasks without complete retraining, by leveraging shared layers across architectures. It offer tools, insights and a precise methodology to empower stakeholders in designing energy-efficient AI systems.

cs.LG↗

arXiv2026

Structured vs. Unstructured Pruning: An Exponential Gap

The Strong Lottery Ticket Hypothesis (SLTH) states that large, randomly initialized neural networks contain sparse subnetworks capable of approximating a target function at initialization without training, suggesting that pruning alone is sufficient. Pruning methods are typically classified as unstructured, where individual weights can be removed from the network, and structured, where parameters are removed according to specific patterns, as in neuron pruning. Existing theoretical results supporting the SLTH rely almost exclusively on unstructured pruning, showing that logarithmic overparameterization suffices to approximate simple target networks. In contrast, neuron pruning has received limited theoretical attention, despite its practical appeal for direct hardware speedups. In this work, we consider the problem of approximating a single bias-free ReLU neuron by pruning hidden units of a randomly initialized two-layer ReLU network, effectively isolating the intrinsic limitations of neuron pruning. We show that achieving an $\varepsilon$-approximation requires a starting network size of $Ω(1/\varepsilon)$ for neuron pruning, whereas weight pruning succeeds with only $O(\log(1/\varepsilon))$ hidden units, revealing an exponential separation between the two approaches.

cs.LG↗

arXiv2026

DejaVu: A Minimalistic Mechanism for Distributed Plurality Consensus

We study the plurality consensus problem in distributed systems where a population of extremely simple agents, each initially holding one of $k$ opinions, aims to agree on the initially most frequent one. In this setting, $h$-majority is arguably the simplest and most studied protocol, in which each agent samples the opinion of $h$ neighbors uniformly at random and updates its opinion to the most frequent value in the sample. We propose a new, extremely simple mechanism called DéjàVu: an agent queries neighbors until it encounters an opinion for the second time, at which point it updates its own opinion to the duplicate value. This rule does not require agents to maintain counters or estimate frequencies, nor to choose any parameter (such as a sample size $h$); it relies solely on the primitive ability to detect repetition. We provide a rigorous analysis of DéjàVu that relies on several technical ideas of independent interest and demonstrates that it is competitive with $h$-majority and, in some regimes, substantially more communication-efficient, thus yielding a powerful primitive for plurality consensus.

cs.DC↗

arXiv2025

Small is Sufficient: Reducing the World AI Energy Consumption Through Model Selection

The energy consumption and carbon footprint of Artificial Intelligence (AI) have become critical concerns due to rising costs and environmental impacts. In response, a new trend in green AI is emerging, shifting from the "bigger is better" paradigm, which prioritizes large models, to "small is sufficient", emphasizing energy sobriety through smaller, more efficient models. We explore how the AI community can adopt energy sobriety today by focusing on model selection during inference. Model selection consists of choosing the most appropriate model for a given task, a simple and readily applicable method, unlike approaches requiring new hardware or architectures. Our hypothesis is that, as in many industrial activities, marginal utility gains decrease with increasing model size. Thus, applying model selection can significantly reduce energy consumption while maintaining good utility for AI inference. We conduct a systematic study of AI tasks, analyzing their popularity, model size, and efficiency. We examine how the maturity of different tasks and model adoption patterns impact the achievable energy savings, ranging from 1% to 98% for different tasks. Our estimates indicate that applying model selection could reduce AI energy consumption by 27.8%, saving 31.9 TWh worldwide in 2025 - equivalent to the annual output of five nuclear power reactors.

cs.CY↗

arXiv2025

Attribute Inference Attacks for Federated Regression Tasks

Federated Learning (FL) enables multiple clients, such as mobile phones and IoT devices, to collaboratively train a global machine learning model while keeping their data localized. However, recent studies have revealed that the training phase of FL is vulnerable to reconstruction attacks, such as attribute inference attacks (AIA), where adversaries exploit exchanged messages and auxiliary public information to uncover sensitive attributes of targeted clients. While these attacks have been extensively studied in the context of classification tasks, their impact on regression tasks remains largely unexplored. In this paper, we address this gap by proposing novel model-based AIAs specifically designed for regression tasks in FL environments. Our approach considers scenarios where adversaries can either eavesdrop on exchanged messages or directly interfere with the training process. We benchmark our proposed attacks against state-of-the-art methods using real-world datasets. The results demonstrate a significant increase in reconstruction accuracy, particularly in heterogeneous client datasets, a common scenario in FL. The efficacy of our model-based AIAs makes them better candidates for empirically quantifying privacy leakage for federated regression tasks.

cs.LG↗

arXiv2024

On the Sparsity of the Strong Lottery Ticket Hypothesis

Considerable research efforts have recently been made to show that a random neural network $N$ contains subnetworks capable of accurately approximating any given neural network that is sufficiently smaller than $N$, without any training. This line of research, known as the Strong Lottery Ticket Hypothesis (SLTH), was originally motivated by the weaker Lottery Ticket Hypothesis, which states that a sufficiently large random neural network $N$ contains \emph{sparse} subnetworks that can be trained efficiently to achieve performance comparable to that of training the entire network $N$. Despite its original motivation, results on the SLTH have so far not provided any guarantee on the size of subnetworks. Such limitation is due to the nature of the main technical tool leveraged by these results, the Random Subset Sum (RSS) Problem. Informally, the RSS Problem asks how large a random i.i.d. sample $Ω$ should be so that we are able to approximate any number in $[-1,1]$, up to an error of $ ε$, as the sum of a suitable subset of $Ω$. We provide the first proof of the SLTH in classical settings, such as dense and equivariant networks, with guarantees on the sparsity of the subnetworks. Central to our results, is the proof of an essentially tight bound on the Random Fixed-Size Subset Sum Problem (RFSS), a variant of the RSS Problem in which we only ask for subsets of a given size, which is of independent interest.

stat.ML↗

arXiv2023

Preferential attachment hypergraph with vertex deactivation

In the field of complex networks, hypergraph models have so far received significantly less attention than graphs. However, many real-life networks feature multiary relations (co-authorship, protein reactions) may therefore be modeled way better by hypergraphs. Also, a recent study by Broido and Clauset suggests that a power-law degree distribution is not as ubiquitous in the natural systems as it was thought so far. They experimentally confirm that a majority of networks (56% of around 1000 networks that undergone the test) favor a power-law with an exponential cutoff over other distributions. We address the two above observations by introducing a preferential attachment hypergraph model which allows for vertex deactivations. The phenomenon of vertex deactivations is rare in existing theoretical models and omnipresent in real-life scenarios (social network accounts which are not maintained forever, collaboration networks in which people retire, technological networks in which devices break down). We prove that the degree distribution of the proposed model follows a power-law with an exponential cutoff. We also check experimentally that a Scopus collaboration network has the same characteristic. We believe that our model will predict well the behavior of systems from a variety of domains.

cs.DM↗

arXiv2023

Revisiting the Random Subset Sum problem

The average properties of the well-known Subset Sum Problem can be studied by the means of its randomised version, where we are given a target value $z$, random variables $X_1, \ldots, X_n$, and an error parameter $\varepsilon > 0$, and we seek a subset of the $X_i$s whose sum approximates $z$ up to error $\varepsilon$. In this setup, it has been shown that, under mild assumptions on the distribution of the random variables, a sample of size $\mathcal{O}(\log(1/\varepsilon))$ suffices to obtain, with high probability, approximations for all values in $[-1/2, 1/2]$. Recently, this result has been rediscovered outside the algorithms community, enabling meaningful progress in other fields. In this work we present an alternative proof for this theorem, with a more direct approach and resourcing to more elementary tools.

math.PR↗

arXiv2021

Preferential attachment hypergraph with high modularity

Numerous works have been proposed to generate random graphs preserving the same properties as real-life large scale networks. However, many real networks are better represented by hypergraphs. Few models for generating random hypergraphs exist and no general model allows to both preserve a power-law degree distribution and a high modularity indicating the presence of communities. We present a dynamic preferential attachment hypergraph model which features partition into communities. We prove that its degree distribution follows a power-law and we give theoretical lower bounds for its modularity. We compare its characteristics with a real-life co-authorship network and show that our model achieves good performances. We believe that our hypergraph model will be an interesting tool that may be used in many research domains in order to reflect better real-life phenomena.

cs.SI↗

arXiv2020

A Random Growth Model with any Real or Theoretical Degree Distribution

The degree distributions of complex networks are usually considered to be power law. However, it is not the case for a large number of them. We thus propose a new model able to build random growing networks with (almost) any wanted degree distribution. The degree distribution can either be theoretical or extracted from a real-world network. The main idea is to invert the recurrence equation commonly used to compute the degree distribution in order to find a convenient attachment function for node connections - commonly chosen as linear. We compute this attachment function for some classical distributions, as the power-law, broken power-law, geometric and Poisson distributions. We also use the model on an undirected version of the Twitter network, for which the degree distribution has an unusual shape. We finally show that the divergence of chosen attachment functions is heavily links to the heavy-tailed property of the obtained degree distributions.

cs.SI↗

arXiv2020

Interest Clustering Coefficient: a New Metric for Directed Networks like Twitter

We study here the clustering of directed social graphs. The clustering coefficient has been introduced to capture the social phenomena that a friend of a friend tends to be my friend. This metric has been widely studied and has shown to be of great interest to describe the characteristics of a social graph. In fact, the clustering coefficient is adapted for a graph in which the links are undirected, such as friendship links (Facebook) or professional links (LinkedIn). For a graph in which links are directed from a source of information to a consumer of information, it is no more adequate. We show that former studies have missed much of the information contained in the directed part of such graphs. We thus introduce a new metric to measure the clustering of a directed social graph with interest links, namely the interest clustering coefficient. We compute it (exactly and using sampling methods) on a very large social graph, a Twitter snapshot with 505 million users and 23 billion links. We additionally provide the values of the formerly introduced directed and undirected metrics, a first on such a large snapshot. We exhibit that the interest clustering coefficient is larger than classic directed clustering coefficients introduced in the literature. This shows the relevancy of the metric to capture the informational aspects of directed graphs.

cs.SI↗

arXiv2014

Energy Efficient Routing by Switching-Off Network Interfaces

Several studies exhibit that the traffic load of the routers only has a small influence on their energy consumption. Hence, the power consumption in networks is strongly related to the number of active network elements, such as interfaces, line cards, base chassis,... The goal thus is to find a routing that minimizes the (weighted) number of active network elements used when routing. In this paper, we consider a simplified architecture where a connection between two routers is represented as a link joining two network interfaces. When a connection is not used, both network interfaces can be turned off. Therefore, in order to reduce power consumption, the goal is to find the routing that minimizes the number of used links while satisfying all the demands. We first define formally the problem and we model it as an integer linear program. Then, we prove that this problem is not in APX, that is there is no polynomial-time constant-factor approximation algorithm. We propose a heuristic algorithm for this problem and we also prove some negative results about basic greedy and probabilistic algorithms. Thus we present a study on specific topologies, such as trees, grids and complete graphs, that provide bounds and results useful for real topologies. We then exhibit the gain in terms of number of network interfaces (leading to a global reduction of approximately 33 MWh for a medium-sized backbone network) for a set of existing network topologies: we see that for almost all topologies more than one third of the network interfaces can be spared for usual ranges of operation. Finally, we discuss the impact of energy efficient routing on the stretch factor and on fault tolerance.

cs.NI↗