HAL-MINES ParisTech
Not a member yet
27348 research outputs found
Sort by
Multilingual Segmentation Dataset (medieval states of vernacular languages)
No full textInternational audienc
Heterogeneity in the productivity of French construction firms: a multilevel analysis
No full textInternational audienceAbstract This paper examines the determinants of productivity heterogeneity among construction firms, integrating both firm-level and local contextual factors. Using a longitudinal dataset of 78,598 firms over 2009–2019, we estimate total factor productivity (TFP) and use a multilevel model across 287 labour market areas in metropolitan France. This approach allows us to separate firm-specific from location-specific effects and explains how the productivity heterogeneity can be attributed to each level. Our results show that firm characteristics significantly affect TFP. Local conditions also play a key role: higher unemployment reduces productivity, whereas higher employment density and median income enhance it. This study contributes by applying a multilevel framework to the construction sector, using longitudinal data for precise estimation, and employing TFP rather than traditional labour productivity as the performance measure. Our findings highlight the joint influence of firm and regional factors on productivity, providing insights for policymakers and managers aiming to improve firm performance and regional economic outcomes in the construction sector. Our results remain robust across different firm size categories, alternative TFP measures, French construction sector sub-sectors and Mundlak's (1978) approach to correcting for heterogeneity bias
IA et cancer de la thyroïde : demain, la fin des traitements standardisés ?
No full textInternational audienceEn couplant intelligence artificielle et modélisation mathématique, un consortium français d’équipes de recherche développe un outil pour, à terme, administrer, en fonction du profil de chaque patient, les doses d’iode radioactif utilisées pour traiter certains types de cancers de la thyroïde. Mais le chemin est encore long avant de voir émerger cette médecine quantitative, prédictive et personnalisée
Des entrepôts de recherche à l’épreuve de l’avenir: les sept piliers d’une infrastructure de recherche robuste, interopérable et compatible avec l’IA
No full textLe GT Interopérabilité & réseaux de Couperin a élaboré ce poster consacré aux enjeux d’interopérabilité, de durabilité et de robustesse des entrepôts de recherche.Intitulé IMPACT-R, ce poster propose une synthèse claire et opérationnelle des sept piliers d’une infrastructure d’entrepôt pérenne, interopérable et compatible avec les évolutions technologiques, notamment liées à l’IA. Il met en lumière, entre autres : la conformité aux principes FAIR (métadonnées normalisées, identifiants pérennes), l’importance des standards ouverts, des API et des intégrations avec les systèmes d’information de la recherche, le rôle central des entrepôts dans les politiques institutionnelles de science ouverte, la nécessité d’un financement pérenne, de compétences professionnelles renforcées et d’une inscription active dans les réseaux nationaux et internationaux, ainsi que les enjeux de sécurité, de fiabilité et de protection à long terme des données.Fruit d’un travail collectif du GT, ce poster s’inspire des réflexions portées au niveau international par OpenAIRE, LIBER, SPARC et COAR, et vise à outiller les établissements dans leurs choix stratégiques autour des entrepôts de recherche.International audienc
AI-Driven Multi-Agent System for Autonomous Mining Operation Centers
No full textInternational audienceAS mining operations enter the era of Industry 4.0, traditional Remote Operation Centers remain constrained by reactive workflows and limited automation. This work introduces EI2ROC, a next-generation agentic AI framework that transforms mining IROCs into proactive, semi-autonomous decision engines. By orchestrating specialized agents for operational intelligence (VDT, OEE, SIC) and deep-learning-based production forecasting, EI2ROC enables continuous diagnostic insight, risk-aware capacity prediction, and automated corrective actions. Early results indicate a potential 70-80% reduction in manual monitoring and significant performance gains through anticipatory control. Designed as a modular, scalable architecture, EI2ROC offers a transferable blueprint for autonomous, data-driven operations across mining and other asset-intensive Industry 4.0 domains.À mesure que les opérations minières entrent dans l'ère de l'Industrie 4.0, les Centres d'Opération à Distance traditionnels restent contraints par des flux de travail réactifs et une automatisation limitée. Ce travail présente EI2ROC, un cadre d'IA agentique de nouvelle génération qui transforme les IROCs miniers en moteurs de décision proactifs et semi-autonomes. En orchestrant des agents spécialisés pour l'intelligence opérationnelle (VDT, OEE, SIC) et la prévision de production par apprentissage profond, EI2ROC permet un diagnostic continu, une prédiction de capacité tenant compte des risques, et des actions correctives automatisées. Les premiers résultats indiquent une réduction potentielle de 70 à 80 % de la surveillance manuelle et des gains de performance significatifs grâce au contrôle anticipatif. Conçu comme une architecture modulaire et évolutive, EI2ROC offre un modèle transférable pour des opérations autonomes et pilotées par les données, applicable au secteur minier et à d'autres domaines industriels à forte intensité d'actifs relevant de l'Industrie 4.0
Constraints on gravitational waves from the 2024 Vela pulsar glitch
No full textInternational audienceAmong known neutron stars, the Vela pulsar is one of the best targets for gravitational-wave searches. It is also one of the most prolific in terms of glitches, sudden frequency changes in a pulsar's rotation. Such glitches could cause a variety of transient gravitational-wave signals. Here we search for signals associated with a Vela glitch on 29 April 2024 in data of the two LIGO detectors from the fourth LIGO--Virgo--KAGRA observing run. We search both for seconds-scale burst-like emission, primarily from fundamental (f-)mode oscillations, and for longer quasi-monochromatic transients up to four months in duration, primarily from quasi-static quadrupolar deformations. We find no significant detection candidates, but for the first time we set direct observational upper limits on gravitational strain amplitude that are stricter than what can be indirectly inferred from the overall glitch energy scale. We discuss the short- and long-duration observational constraints in the context of specific emission models. These results demonstrate the potential of gravitational-wave probes of glitching pulsars as detector sensitivity continues to improve
CompNO: A Novel Foundation Model approach for solving Partial Differential Equations
No full textInternational audiencePartial differential equations (PDEs) govern a wide range of physical phenomena, but their numerical solution remains computationally demanding, especially when repeated simulations are required across many parameter settings. Recent Scientific Foundation Models (SFMs) aim to alleviate this cost by learning universal surrogates from large collections of simulated systems, yet they typically rely on monolithic architectures with limited interpretability and high pretraining expense. In this work we introduce Compositional Neural Operators (CompNO), a compositional neural operator framework for parametric PDEs. Instead of pretraining a single large model on heterogeneous data, CompNO first learns a library of Foundation Blocks, where each block is a parametric Fourier neural operator specialized to a fundamental differential operator (e.g. convection, diffusion, nonlinear convection). These blocks are then assembled, via lightweight Adaptation Blocks, into task-specific solvers that approximate the temporal evolution operator for target PDEs. A dedicated boundary-condition operator further enforces Dirichlet constraints exactly at inference time. We validate CompNO on one-dimensional convection, diffusion, convection-diffusion and Burgers' equations from the PDEBench suite. The proposed framework achieves lower relative L2 error than strong baselines (PFNO, PDEFormer and in-context learning based models) on linear parametric systems, while remaining competitive on nonlinear Burgers' flows. The model maintains exact boundary satisfaction with zero loss at domain boundaries, and exhibits robust generalization across a broad range of Péclet and Reynolds numbers. These results demonstrate that compositional neural operators provide a scalable and physically interpretable pathway towards foundation models for PDEs
Long-range minimal models
No full textInternational audienceWe study a class of nonlocal conformal field theories in two dimensions which are obtained as deformations of the Virasoro minimal models. The construction proceeds by coupling a relevant primary operator of the -th minimal model to a generalized free field, in such a way that the interaction term has scaling dimension . Flowing to the infrared, we reach a new class of CFTs that we call long-range minimal models. In the case , the resulting line of fixed points, parametrized by , can be studied using two perturbative expansions with different regimes of validity, one near the mean-field theory end, and one close to the long-range to short-range crossover. This is due to a straightforward generalization of an infrared duality which was proposed for the long-range Ising model () in 2017. We find that the large- limit is problematic in both perturbative regimes, hence nonperturbative methods will be required in the intermediate range for all values of . For the models based on , the situation is rather different. In this case, only one perturbative expansion is known but it is well behaved at large . We confirm this with a computation of infinitely many anomalous dimensions at two loops. Their large- limits are obtained from both numerical extrapolations and a method we develop which carries out conformal perturbation theory using Mellin amplitudes. For minimal models, these can be accessed from the Coulomb gas representations of the correlators. This method reveals analytic expressions for some integrals in conformal perturbation theory which were previously only known numerically
Oncogene-driven advocacy: Collective expertise and therapeutic actionability
No full textInternational audienceThis paper is intended as a contribution to the social science literature on Patient Advocacy Groups (PAGs). It examines the recent emergence and development of cancer patient organizations that self-define as “oncogene-focused,” that is, as centered on tumor-driving genetic mutations that offer novel therapeutic opportunities for tumors to be targeted by precision drugs. Drawing on qualitative methods including interviews with representatives of oncogene focused PAGs, analysis of the groups’ publications (biomedical journals, eNewsletters), websites, and social media activity, the paper explores the characteristics of these PAGs’ forms of activism. It shows that their common denominator is a focus on patient survival. This shared goal translates into a form of activism that centers on therapeutic actionability , that is, a set of initiatives aiming at the articulation of research, clinical trials, and care to improve the patients’ quality of life and maximize survival thanks to awareness of and access to the latest therapies. Beyond individual differences between PAGs, we observe the increasingly seamless entanglement of their activities. Their mutually supportive interventions result in the establishment of an ecosystemic form of activism that also succeeds in mobilizing clinicians and researchers at the increasingly porous interface between research and care
Le Grand Paris Express aux prises avec le calcul économique
No full textInternational audienceCet article propose une analyse de l’approche socio-économique développée par la Société du Grand Paris pour justifier de l’utilité du Grand Paris Express (GPE) lors des procédures d’enquête publique et de contre-expertise. Cette approche a en effet suscité une controverse mobilisant de nombreux acteurs académiques et associatifs ayant dénoncé une surévaluation de sa rentabilité au service d’un aménagement reposant sur l’idée d’« effets structurants » des réseaux de transport. Il montre comment, face à une rentabilité socio-économique initialement défavorable en raison des techniques d’évaluation en vigueur et de la morphologie du GPE, la SGP est parvenue à imposer une nouvelle méthode assise sur l’idée d’une meilleure intégration des interactions entre réseau et territoire dans le calcul économique. En étudiant les thèses et les modèles développés par les experts du service d’études économiques de la SGP ainsi que les controverses qu’elles ont suscitées dans différentes arènes, l’article propose ensuite une réflexion sur les problèmes de confiance posés par les modèles Land Use Transport Interaction (LUTI) et sur le type de vision du territoire que ce calcul économique incarne par sa méthode