HAL-Ecole des Ponts ParisTech
Not a member yet
42743 research outputs found
Sort by
AI-boosted rare event sampling to characterize extreme weather
No full textAssessing the frequency and intensity of extreme weather events, and understanding how climate change affects them, is crucial for developing effective adaptation and mitigation strategies. However, observational datasets are too short and physics-based global climate models (GCMs) are too computationally expensive to obtain robust statistics for the rarest, yet most impactful, extreme events. AI-based emulators have shown promise for predictions at weather and even climate timescales, but they struggle on extreme events with few or no examples in their training dataset. Rare event sampling (RES) algorithms have previously demonstrated success for some extreme events, but their performance depends critically on a hard-to-identify "score function", which guides efficient sampling by a GCM. Here, we develop a novel algorithm, AI+RES, which uses ensemble forecasts of an AI weather emulator as the score function to guide highly efficient resampling of the GCM and generate robust (physics-based) extreme weather statistics and associated dynamics at 30-300x lower cost. We demonstrate AI+RES on mid-latitude heatwaves, a challenging test case requiring a score function with predictive skill many days in advance. AI+RES, which synergistically integrates AI, RES, and GCMs, offers a powerful, scalable tool for studying extreme events in climate science, as well as other disciplines in science and engineering where rare events and AI emulators are active areas of research
When Gender Kicks in: an Experimental Study of Work from Home and Attitudes to Household Work and Childcare
No full textWe study how working from home links to gendered attitudes about household work and childcare. Using a vignette experiment embedded in a regular Dutch population representative survey, we randomly vary the gender of the partner working from home in a hypothetical dualearner couple. When presented with various routine and emergency chores, respondents, on average, agree that the partner working from home should execute them, and the extent of agreement is significantly larger when the vignette randomly depicts a man, rather than a woman, working from home. These differences in respondents' gendered expectations around performing chores are not statistically significant in the baseline scenario where no partner works from home. All in all, the evidence gathered indicates that Work from Home may blast rather than boost gender norms around household work and childcare
Vers de nouveaux indicateurs rapides et à haute fréquence pour suivre les concentrations en E. coli par spectrométrie de fluorescence dans la Seine pour la gestion active de zones de baignade
No full textEtude parue dans le numéro 1&2 de TSM en 2025International audienceLes deux dernières décennies ont vu s’intensifier le développement de nouveaux appareils de métrologie à haute fréquence pour suivre les paramètres biophysicochimiques des eaux de surface. Les sondes optiques, en particulier les sondes de fluorescence, occupent une place prépondérante dans cette démarche d’intégration de la mesure haute fréquence aux suivis environnementaux. L’estimation en continu de la concentration en bactéries indicatrices fécales (BIF) fait partie des applications émergentes associées aux sondes optiques. Ainsi, une campagne de prélèvements et d’analyses a été organisée en plein coeur de Paris à la fin de l’été 2023. Elle a permis de recueillir 135 échantillons d’eau de Seine permettant de confronter des mesures acquises par spectrométrie de fluorescence à des concentrations de BIF obtenues par les méthodes normées de référence (NF EN ISO 9308-3 et NF-EN ISO 7899-1). L’extraction de 325 variables explicatives issues des spectres de fluorescence a permis de construire de nombreux modèles de prédiction en utilisant notamment des algorithmes de Machine Learning. Les meilleurs d’entre eux (R² = 0,77) démontrent le potentiel de la spectrométrie de fluorescence pour estimer la concentration en Escherichia coli observée en Seine, à Paris. La méthodologie mise en oeuvre à l’occasion de ce travail pourra par la suite servir de référence pour calibrer de nouveaux modèles adaptés spécifiquement à d’autres sites de baignade. La spectrométrie de fluorescence, via l’utilisation de la sonde de fluorescence Fluocopée, pourra alors constituer un outil de surveillance in situ et à haute fréquence pour garantir en continu la santé des baigneurs
RUBIK: A Structured Benchmark for Image Matching across Geometric Challenges
No full textInternational audienceCamera pose estimation is crucial for many computer vision applications, yet existing benchmarks offer limited insight into method limitations across different geometric challenges. We introduce RUBIK, a novel benchmark that systematically evaluates image matching methods across well-defined geometric difficulty levels. Using three complementary criteria - overlap, scale ratio, and viewpoint angle - we organize 16.5K image pairs from nuScenes into 33 difficulty levels. Our comprehensive evaluation of 14 methods reveals that while recent detector-free approaches achieve the best performance (>47% success rate), they come with significant computational overhead compared to detector-based methods (150-600ms vs. 40-70ms). Even the best performing method succeeds on only 54.8% of the pairs, highlighting substantial room for improvement, particularly in challenging scenarios combining low overlap, large scale differences, and extreme viewpoint changes. Benchmark will be made publicly available
La fiscalité verte, une mise en œuvre inégale
No full textNational audienceLa fiscalité verte, une mise en œuvre inégale
Correcting for water vapor diffusion in air bag samples for isotope composition analysis: case study with drone-collected samples
No full textInternational audienceMass spectrometry and laser spectroscopy have been widely employed for precise water vapor isotope measurements. Nevertheless, these techniques are limited by logistical challenges in fieldwork, consequently constraining the temporal and spatial resolution of measurements. Specifically, water vapor isotope measurements are primarily limited to near-surface levels, while measurements in the free troposphere are notably scarce. Portable sampling devices, such as air bags and glass bottles, have therefore become necessary alternatives for collecting, storing, and transporting gaseous samples in diverse environments prior to analysis with less portable instruments. In drone-based high-altitude vapor sampling, air bags are preferred for their lighter weight and greater flexibility compared to glass bottles. Nevertheless, they present specific challenges, such as potential sample contamination and isotopic fractionation during storage, primarily due to the inherent permeability of air bags. Here, we developed a theoretical model for water vapor diffusion through the sampling bag surface, with parameters calibrated through laboratory experiments. This model enables the reconstruction of the initial isotopic composition of sampled vapor based on measurements obtained within the bag and from the surrounding environment. This diffusion model underwent rigorous validation through experiments conducted under varying humidity and isotopic composition differences between the inside and outside of the air bag, confirming its reliability. We applied this correction method to air samples collected at various pressures up to the upper troposphere using an air bag-mounted drone that we developed, thereby estimating the initial isotopic composition and uncertainty based on our observations. Our correction method enhances the reliability and applicability of water vapor isotope observations conducted using drones equipped with air bags, and provides a detailed assessment of all potential sources of error and quantifies the uncertainty range of the observations. This approach leverages the strengths of drone-based air bag sampling while mitigating its limitations, thus facilitating the convenient collection of isotopic data throughout the troposphere
eHPWAS'25: Thirteenth International workshop on e-health pervasive wireless applications and services
No full textInternational audienc
Human Manure as Activism: Composting Excrement as an Alternative Approach to Soil Fertilization
No full textInternational audienceFollowing a period of intense interest in the use of human excrement as fertilizer in nineteenth-century Europe, this type of manure was largely sidelined by the rise of synthetic fertilizers. However, interest in human excreta as manure has seen a resurgence in recent decades thanks to the growing use of composting toilets at the domestic level, particularly in France. In this chapter, we explore how the practice of returning human excrement to the soil relates to agricultural production at different points in recent history. Specifically, we examine how these practices provide a basis for a political and economic critique of the prevailing extractive model and serve as a means of inquiry into the (re)production of soil fertility
Efficient Numerical Integration in Reproducing Kernel Hilbert Spaces via Leverage Scores Sampling
No full textInternational audienceIn this work we consider the problem of numerical integration, i.e., approximating integrals with respect to a target probability measure using only pointwise evaluations of the integrand. We focus on the setting in which the target distribution is only accessible through a set of n i.i.d. observations, and the integrand belongs to a reproducing kernel Hilbert space. We propose an efficient procedure which exploits a small i.i.d. random subset of m < n samples drawn either uniformly or using approximate leverage scores from the initial observations. Our main result is an upper bound on the approximation error of this procedure for both sampling strategies. It yields sufficient conditions on the subsample size to recover the standard (optimal) n^{-1/2} rate while reducing drastically the number of functions evaluations-and thus the overall computational cost. Moreover, we obtain rates with respect to the number m of evaluations of the integrand which adapt to its smoothness, and match known optimal rates for instance for Sobolev spaces. We illustrate our theoretical findings with numerical experiments on real datasets, which highlight the attractive efficiency-accuracy tradeoff of our method compared to existing randomized and greedy quadrature methods. We note that, the problem of numerical integration in RKHS amounts to designing a discrete approximation of the kernel mean embedding of the target distribution. As a consequence, direct applications of our results also include the efficient computation of maximum mean discrepancies between distributions and the design of efficient kernel-based tests
Harmonised boundary layer wind profile dataset from six ground-based Doppler wind lidars in a transect across Paris, France
No full textInternational audienceDoppler wind lidars (DWL) offer high-resolution wind profile measurements that are valuable for understanding atmospheric boundary layer (ABL) dynamics. Here six ground-based DWL, deployed in a multi-institutional effort along a 40 km transect through the centre of Paris (France), are used to retrieve horizontal wind speed and direction through the ABL at 18–25 m vertical and 1–60 min temporal resolution. Data are available for June 2022–March 2024 (three DWL) and two Intensive Observation Periods (six DWL) across 9 weeks in September 2023–December 2023. Data from all sensors are harmonised in terms of quality control, file format, as well as temporal and vertical resolutions. The quality of this DWL dataset is evaluated against in-situ measurements at the Eiffel Tower and radiosonde profiles. This unique, spatially dense, open dataset will allow urban boundary layer dynamics to be explored in process-studies, and is further valuable for the evaluation of high-resolution weather, climate, inverse and air pollution models that resolve city-scale processes. The dataset is available at https://doi.org/10.5281/zenodo.14761503 (Morrison et al., 2025)