HAL Portal IMT Nord Europe
Not a member yet
6687 research outputs found
Sort by
Quenched large deviations for Monte Carlo integration with Coulomb gases
No full textGibbs measures, such as Coulomb gases, are popular in modelling systems of interacting particles. Recently, we proposed to use Gibbs measures as randomized numerical integration algorithms with respect to a target measure π on R d , following the heuristics that repulsiveness between particles should help reduce integration errors. A major issue in this approach is to tune the interaction kernel and confining potential of the Gibbs measure, so that the equilibrium measure of the system is the target distribution π. Doing so usually requires another Monte Carlo approximation of the potential, i.e. the integral of the interaction kernel with respect to π. Using the methodology of large deviations from Garcia-Zelada (2019), we show that a random approximation of the potential preserves the fast large deviation principle that guarantees the proposed integration algorithm to outperform independent or Markov quadratures. For non-singular interaction kernels, we make minimal assumptions on this random approximation, which can be the result of a computationally cheap Monte Carlo preprocessing. For the Coulomb interaction kernel, we need the approximation to be based on another Gibbs measure, and we prove in passing a control on the uniform convergence of the approximation of the potential
Evaluation of Multi-Relay D2D Communication for Cooperative SLAM in Metaverse Context
No full textInternational audienceIn conventional deployments, infrastructure-based communication routes all traffic through the gNB and corenetwork, providing centralized orchestration and global persistence. However, in indoor scenarios such as museums, where users are physically co-located, this mode may introduce unnecessary delay and backhaul congestion. Device-to-Device (D2D) relaying emerges as a promising alternative, enabling nearby devices to exchange data directly, either in single-hop or multi-hop configurations. While existing studies on D2D multicast mainly focus on resource allocation for VR content sharing, the role of relay-based D2D multicast in supporting latency-sensitive Metaverse services has not been fully addressed. To this end, this paper extends the Simu5G platform with loosely coupled relay functionality and provides a comparative analysis of infrastructure and D2D multicast communication modes for cooperative Simultaneous Localization And Mapping (SLAM) in an indoor museum scenario. Simulation results show that relay-based D2D multicast can serve as a complementary solution to infrastructure-based communication, particularly when museumvisitors move toward the cell edge
Experimental investigation of mechanical behavior and thermal damage of hot dry rock exposing to different cooling conditions
No full textInternational audienc
Investigation numérique des effets de seuil dans l'érosion interne des sols granulaires par couplage DEM-DFM
No full textInternational audienc
Évaluation des sols contaminés en cours de renaturation à l’aide d’indicateurs de la fonctionnalité et de l’écotoxicité des sols pollués aux métaux
No full textInternational audienceL’écosystème constitué par le continuum sol-plante-eau peut être exposé à des situations de pollution, notamment aux métaux. De nouvelles utilisations des sols pollués peuvent être envisagées afin de les protéger et de restaurer voire d’améliorer leurs fonctions tout en réduisant l’exposition aux polluants. Dans ce contexte, des approches de réhabilitation écologique qui prennent en compte le degré de dégradation de l’écosystème, les utilisations futures des terres, le caractère abordable des solutions et l’impact du changement climatique peuvent être pertinentes. Ainsi, à travers 2 projets en cours (REVE, REECOL), plusieurs couvertures végétales sur plusieurs sites contaminés par des métaux seront suivies. De plus, divers indicateurs physico-chimiques, biologiques et écotoxicologiques seront mesurés pour caractériser la qualité du sol en relation avec ces couverts végétaux ainsi que la qualité et la santé de ces derniers. Les meilleures stratégies pour réduire l’exposition aux métaux tout en offrant d’autres avantages sur les fonctions du sol, l’amélioration de la biodiversité et les services écosystémiques seront étudiées. Les résultats présentés ici se concentrent exclusivement sur le projet REVE
Surrogate-based sensitivity analysis on the landscape evolution model of an elementary agricultural catchment
No full textInternational audienc
Analysis of source regions and transport pathways of sub-micron aerosol components in Europe
No full textInternational audienceIt is important to study aerosols and their origins, as they pose various negative health and environmental impacts. In this study, we combined year-long datasets from 15 different countries with Trajectory Statistical Methods (TSMs) on such a comprehensive scale for the first time. We found possible source regions and seasonal variations of various particulate matter (PM) components, including total organic aerosol (OA), biomass burning OA (BBOA), oxygenated OA (OOA), ammonium (NH4), nitrate (NO3), and sulphate (SO4) in Europe. We found that for all of the studied components, Eastern Europe was among the highest contributors. For NO3, other important source regions were northern France and the Benelux, while for SO4, there were significant contributions from the Mediterranean region. We also compared our measurementbased model with simulated concentrations of an atmospheric chemistry transport model (CAMx).We observed a satisfactory agreement in regions where we had sufficient coverage with air pollution monitoring stations. The main deviations for OA were found around the Po Valley, where CAMx consistently estimated higher concentrations, while the TSM analysis did not highlight it as a hot spot because long-term monitoring data sets in this region are lacking. CAMx also underestimated the concentrations around Poland, mainly from residential burning. Our results provide opportunities to refine European emission inventories and deliver valuable information on long-range transported air pollutants. It suggests that policies mitigating air pollution in Eastern Europe and the Benelux could help improve overall air quality in entire Europe more efficiently.</div
Unleashing the Power of Gradual Patterns for Explainable AI
No full textEnsemble models and deep neural networks (DNNs) demonstrate excellent results in classification tasks. However, their "black-box" nature prevents their widespread deployment and use in critical fields such as health. Explainable AI is a field to make black-box models more understandable by humans. In the literature, predictions are explained mostly in the form of feature attributions or counterfactuals based on neighborhood generated randomly or using genetics algorithms or with expert knowledge. In this paper, we show how gradual patterns can be used to generate more plausible neighborhoods without requiring expert knowledge, producing explanations better adapted to individual instances. We extend our post-hoc explainable AI framework with a comprehensive theoretical analysis and additional experimental results, comparing it with state-of-the-art methods such as LIME, LORE, and SHAP, and discuss practical implementation considerations
Particulate and gas emissions from wildfires in the southern Amazon from GOES-16 fire radiative power retrievals
No full textInternational audienceThis work uses geostationary satellite GOES-16 data from 2020 to 2022 to study fire patterns and emissions in the southern Amazon. The Fire/Hotspot Characterization algorithm processes GOES-16's Advanced Baseline Imager data, providing insights into fire dynamics with an unparalleled temporal resolution of 10 minutes. This approach allows for retrieving accurate temporal evolution of fire radiative power (FRP) and emission estimates compared to previous efforts based on polar satellites