Archives ouvertes Hal IMT Atlantique
Not a member yet
20414 research outputs found
Sort by
Comprendre les désordres de l’information : état de l’art des mécanismes, vulnérabilités et réponses dans l’environnement informationnel numérique
No full textMisinformation is one of the most serious threats to contemporary democracies, yet its underlying mechanisms are poorly understood. This report provides a comprehensive overview of the various forms of misinformation and how it spreads in digital environments. It challenges oversimplified explanations that blame gullibility or the deliberate circulation of false content as the sole cause.The analysis reveals that misinformation is a diverse phenomenon whose impact depends more on the context in which it circulates than on its falsity or sophistication. The report points to the structural role played by sociotechnical architectures—including algorithmic dynamics, media relay effects, and coordinated influence operations—in giving certain content disproportionate visibility, regardless of how broadly it is actually endorsed. Informational vulnerabilities are relational and context-dependent, arising from the interplay of ordinary cognitive tendencies, emotional dynamics, and digital environments designed to capture attention.In light of these challenges, the report advocates a comprehensive strategy that combines systemic countermeasures to diminish the spread and financial gain from deceptive content with individual-level interventions integrated into platform architectures. The overarching goal is a pragmatic harm-reduction framework that balances the protection of free expression with the demands of democratic deliberation.La mésinformation constitue l'une des principales menaces pesant sur les démocraties contemporaines, sans que ses mécanismes profonds ne soient toujours bien compris. Ce rapport propose un état de l'art synthétique sur ses définitions, ses formes et ses dynamiques de diffusion dans l'environnement numérique, en s'écartant des lectures réductrices qui l'imputent à la seule crédulité individuelle ou à la prolifération de contenus délibérément faux.L'analyse montre que la mésinformation est un phénomène hétérogène dont l'impact dépend moins de la fausseté ou de la sophistication des contenus que du contexte dans lequel ils circulent et sont reçus. Elle met en évidence le rôle structurant des architectures socio-techniques, notamment les dynamiques algorithmiques, les relais médiatiques et les stratégies coordonnées, dans la sur-visibilité de certains contenus, indépendamment de leur adhésion réelle dans la population. Les vulnérabilités informationnelles y apparaissent comme relationnelles et contextuelles, résultant de l'interaction entre mécanismes cognitifs ordinaires, dynamiques émotionnelles et environnements numériques orientés vers la captation de l'attention.Face à ces constats, ce rapport plaide pour une approche intégrée, articulant contre-mesures systémiques agissant sur l'amplification et la rentabilité des contenus trompeurs et interventions individuelles déployées au sein même des plateformes. Il entend ainsi soutenir une réduction pragmatique des risques, compatible avec la préservation de la liberté d'expression et les exigences du débat démocratique
Measurement of the p correlation function in pp collisions at TeV
No full textInternational audienceIn this letter, the first measurement of the femtoscopic correlation of protons and hyperons is presented and used to study the p interaction. The measurement is performed with the ALICE detector in high-multiplicity triggered pp collisions at TeV. The hyperons are reconstructed using a missing-mass approach in the decay channel to with , while both and protons are identified using a machine learning approach. These techniques result in a high reconstruction efficiency and purity, which allows the measurement of the p correlation function for the first time. Thanks to the high significance achieved in the p correlation signal, it is possible to discriminate between the predictions of different models of the N interaction and to accomplish a first determination of the p scattering parameters
Search for an eV-scale sterile neutrino with the first six detection units of KM3NeT/ORCA
No full textInternational audienceThe existence of an eV-scale sterile neutrino has been proposed to explain several anomalous experimental results obtained over the course of the past 25 years. The first search for such a sterile neutrino conducted with data from KM3NeT/ORCA -- a water Cherenkov neutrino telescope under construction at the bottom of the Mediterranean Sea -- is reported in this paper. GeV-scale atmospheric neutrino oscillations are measured by reconstructing the energy and arrival direction of up-going neutrinos that have traversed the Earth. This study is based on a data sample containing 5828 neutrino candidates collected with 6 detection units ( of the complete detector), corresponding to an exposure of 433 kton-years. From the expected effect of an eV-scale sterile neutrino on the first standard oscillation maximum, simultaneous constraints are put on the magnitude of the and mixing elements under the assumption eV. The results are compatible with the absence of mixing between active neutrinos and a sterile state, with and at a confidence level. Such constraints are compatible with the results reported by other long-baseline experiments, and indicate that with KM3NeT/ORCA it is possible to bring crucial contributions to sterile neutrino searches in the coming years
Multiplicity dependence of K production in pp collisions at = 13 TeV
No full textInternational audienceThe first results of K(892) production at midrapidity () in pp collisions at TeV as a function of the event multiplicity are presented. The K(892) has been reconstructed via its hadronic decay channel K(892) using the ALICE detector at the LHC. For each multiplicity class the differential transverse momentum () spectrum, the mean transverse momentum , the -integrated yield (d/d), and the ratio of the K(892) to yields are reported. These are consistent with previous K(892) resonance results with a higher level of precision. Comparisons with phenomenological models such as PYTHIA6, PYTHIA8, EPOS-LHC, and DIPSY are also discussed. A first evidence of a significant K(892)/ suppression in pp collisions is observed at a 7 level passing from low to high multiplicity events. The ratios of the -differential yields of K(892) and in high and low multiplicity events are also presented along with their double ratio. For GeV/ this double ratio persists below unity by more than suggesting that the suppression affects mainly low resonances. The measured decreasing trend of the K(892)/ ratio with increasing multiplicity, which in heavy-ion collisions is typically attributed to the rescattering of decay particles of the short-lived resonances, is reproduced by the EPOS-LHC model without the use of hadronic afterburners
Transverse momentum dependent factorisation in the target fragmentation region at small
No full textInternational audienceWe consider the differential cross-section for single-inclusive jet production with transverse momentum in Deep Inelastic Scattering (DIS) at small Bjorken , mediated by a virtual photon with virtuality . Unlike most studies at small , which focus on particle production in the current fragmentation region, we investigate the kinematic regime where the jet is produced in the target fragmentation hemisphere of the Breit frame, and with . For a longitudinally polarised photon, we demonstrate that this cross-section is not power suppressed in and we derive a factorised expression in terms of transverse momentum dependent (TMD) quark and gluon fracture functions. Our formula, valid at next-to-leading order in at small , is akin to the Altarelli-Martinelli identity for the longitudinal DIS structure function. Numerical estimates show that the quark TMD fracture function is the most sensitive to saturation effects in large nuclei
Real-time Instruction-Level Anomaly Detection for Embedded Applications using AI
No full textInternational audienceBare-metal embedded systems, such as ARM Cortex-M4-based devices, are vulnerable to attacks such as buffer overflows due to the lack of operating system protection. This paper presents a novel approach for detecting standard C library functions -such as memcpy, memset, strncat-that are susceptible to such vulnerabilities by analyzing micro-architectural instruction traces.We propose machine learning pipelines, including CNN-, LSTM-, and autoencoder-based detectors. Our approach uses data pre-processing techniques, such as sliding windows with varying stride are employed to optimize classification accuracy.Evaluating the algorithm with 25 custom workloads simulating common weaknesses (e.g., CWE-120, CWE-126) shows 93.89% TPR, 73.19% TNR, 26.81% FPR, and 6.11% FNR.This work advances IoT security by enabling online and real-time vulnerable function identification supporting zero-day attack detection. This goes beyond existing techniques targeting only higher-level platforms.</div
MONET: Modeling and Optimization of neural NEtwork Training from Edge to Data Centers IEEE Publication
No full textWhile hardware-software co-design has significantly improved the efficiency of neural network inference, modeling the training phase remains a critical yet underexplored challenge. Training workloads impose distinct constraints, particularly regarding memory footprint and backpropagation complexity, which existing inference-focused tools fail to capture. This paper introduces MONET, a framework designed to model the training of neural networks on heterogeneous dataflow accelerators. MONET builds upon Stream, an experimentally verified framework that that models the inference of neural networks on heterogeneous dataflow accelerators with layer fusion. Using MONET, we explore the design space of ResNet-18 and a small GPT-2, demonstrating the framework's capability to model training workflows and find better hardware architectures. We then further examine problems that become more complex in neural network training due to the larger design space, such as determining the best layer-fusion configuration. Additionally, we use our framework to find interesting trade-offs in activation checkpointing, with the help of a genetic algorithm. Our findings highlight the importance of a holistic approach to hardwaresoftware co-design for scalable and efficient deep learning deployment
Adaptive Robust Workforce Allocation with Fairness
No full textNational audienceAdaptive Robust Workforce Allocation with Fairnes
Hydrological regime shifts in Sahelian watersheds: an investigation with a simple dynamical model driven by annual precipitation
No full textInternational audienceAbstract. The Sahel, the semi-arid fringe south of the Sahara, experienced severe meteorological droughts in the 1970s–1980s. During and after these droughts, watersheds in the Central Sahel have experienced an increase in the annual runoff coefficient (annual runoff normalized by annual precipitation). We hypothesize that these increases correspond to regime shifts. To investigate the timing of these regime shifts, we introduce a lumped model that represents feedbacks between soil, water and vegetation at the watershed scale and the annual time step. This model relies on runoff coefficient as a constraint for the state variable and precipitation as unique external forcing. Four watersheds (Gorouol, Dargol, Nakanbé and Sirba), with pluri-decennial observations (1950s–2010s), are modeled. For each watershed, one million parameterizations of this model are sampled and run, and an ensemble of one thousand best parameterizations is selected based on observed runoff coefficients. Our results show that this model can reproduce the trend of runoff coefficients. For all watersheds, almost all selected parameterizations from the ensemble are bistable. We define two alternative runoff coefficient regimes (a low and a high regime) by splitting with a threshold the bifurcation diagram of bistable parameterizations. Most selected parameterizations undergo regime shifts: simulated runoff coefficients belong to the low regime in 1965 and to the high regime in 2014. Finally, we find that the year of the regime shift, defined as the year when the number of regime shifts is maximized, was 1971, 1972, 1973, 1983 for the Gorouol, Nakanbé, Dargol and Sirba watershed, respectively. These results were obtained with a parsimonious model which deliberately neglects fine-scale processes of Sahelian hydrology. It would therefore be wise to supplement this analysis with other models – with varying levels of complexity – that also allow regime shifting. Overall, this article proposes simple ideas toward improving the modelling and characterization of hydrological regime shifts
Upper Limits on the Isotropic Gravitational-Wave Background from the first part of LIGO, Virgo, and KAGRA's fourth Observing Run
No full textInternational audienceWe present results from the search for an isotropic gravitational-wave background using Advanced LIGO and Advanced Virgo data from O1 through O4a, the first part of the fourth observing run. This background is the accumulated signal from unresolved sources throughout cosmic history and encodes information about the merger history of compact binaries throughout the Universe, as well as exotic physics and potentially primordial processes from the early cosmos. Our cross-correlation analysis reveals no statistically significant background signal, enabling us to constrain several theoretical scenarios. For compact binary coalescences which approximately follow a 2/3 power-law spectrum, we constrain the fractional energy density to (95% cred.), a factor of 1.7 improvement over previous results. Scale-invariant backgrounds are constrained to , representing a 2.1x sensitivity gain. We also place new limits on gravity theories predicting non-standard polarization modes and confirm that terrestrial magnetic noise sources remain below detection threshold. Combining these spectral limits with population models for GWTC-4, the latest gravitational-wave event catalog, we find our constraints remain above predicted merger backgrounds but are approaching detectability. The joint analysis combining the background limits shown here with the GWTC-4 catalog enables improved inference of the binary black hole merger rate evolution across cosmic time. Employing GWTC-4 inference results and standard modeling choices, we estimate that the total background arising from compact binary coalescences is at 90% confidence, where the largest contribution is due to binary black holes only,