Portail HAL UPEC
Not a member yet
55480 research outputs found
Sort by
A method to estimate absolute odorant concentration of olfactory stimuli
No full textInternational audienceThe accurate quantification and delivery of odorant concentrations remain a significant challenge. Traditional methods estimate stimulus intensity based on the amount of odorant in the source, but this does not reflect the actual concentration sent due to variable evaporation rates and delivery devices. This leads to inconsistencies in stimulus delivery, complicating cross-laboratory comparisons, threshold evaluations, and the replication of natural olfactory conditions in the lab. To address this, we present a model based on mass transfer theory to predict the concentration of odorants delivered by a simple and versatile odor delivery system commonly used in insect electrophysiological experiments. The present model, built with adaptable compartments, accounts for airflow, source size, and the physicochemical properties of odorants. It helps to better design and use odor delivery systems, especially for stimuli required to mimic natural odor environments. Calibration uses known partition coefficients. The model also considers the dynamic shape of odor stimuli, which affects neuronal responses and must be carefully interpreted, especially when using tools like photoionisation detectors (PID). This approach was applied to study the impact of a plant volatile known to activate pheromone-sensitive neurons, (Z)-3-hexenyl acetate, on pheromone detection in Agrotis ipsilon moths. While interference occurs in laboratory conditions at 160 ppb, such concentrations are unlikely in natural settings, suggesting these effects are less relevant ecologically
GloFM: a GLORYS Flow-Matching emulator for spatio-temporal ocean data assimilation
No full textInternational audienceProviding regular and physically consistent predictions of the ocean state is critical for numerous scientific, operational, and societal needs. Observations of the ocean surface are gathered through various remote sensing and in situ instruments, and are typically assimilated into numerical models to reconstruct the ocean state. However, this often involves millions of data points, making it computationally intensive, which suggests deep learning may be a cheaper alternative. Deterministic data-driven approaches typically learn about ocean dynamics from numerical simulations or sparse observational data. However, such methods often lack physical realism in uncertain settings. Due to mode averaging, they produce non-physical or overly simplified states. Generative models offer a promising approach to generating physically realistic ocean states. We present GloFM: a Glorys Flow-Matching emulator for spatio-temporal ocean data assimilation. Our generative model produces coherent estimates of ocean surface fields. GloFM uses flow matching to assimilate observational data for nowcasting of surface currents, sea surface height (SSH), and sea surface temperature (SST). Compared to deterministic regression-based approaches, GloFM demonstrates improved realism metrics, capturing finer-scale variability and more physically plausible ocean states
Séminaire d'initiation aux jeux de rôle sur table en école de management
No full textInternational audienceDans le cadre du projet pédagogique et de recherche en ludopédagogie EdUTeam, nous avons organisé un séminaire d'initiation au jeu de rôle sur table (JdR) durant la semaine de pré-rentrée de la promotion 2025-2026 de la 1 re année de licence Informatique & Management (IAE Paris-Est). Trois objectifs ont guidé notre démarche : proposer un défi collectif aux étudiants qui ne se connaissaient pas encore, aborder de manière concrète le projet EdUTeam, très présent dans les activités pédagogiques de la formation, et faire découvrir le JdR afin de faciliter son déploiement dans de futurs projets ludopédagogiques. En effet, le jeu de rôle, sous toutes ses formes, est une pratique pédagogique largement déployée dans les sciences de gestion et du management. Pour autant, les leviers nécessaires à son déploiement efficace dans un contexte d'enseignement obligatoire nous semblent peu abordés. Cet article tente d'apporter des éléments de réponse. Il détaille les résultats, particulièrement positifs, obtenus grâce à l'administration d'un questionnaire anonyme en fin de semaine ; et ce malgré les obstacles et préjugés initiaux importants. Ils mettent également en lumière des éléments intéressants provenant de ce public de jeunes adultes face à une pratique ludique exigeante, contrairement aux jeux vidéo où les joueurs peuvent s'immerger dans les univers sans aucune préparation
Euler-type approximation for the invariant measure: An abstract framework
No full textWe establish a general framework to study the rate of convergence of a Euler type approximation scheme with decreasing time steps to the invariant measure, for a general class of stochastic systems. The error is measured in general Wasserstein distances, which enables to encompass cases with non global contractivity conditions. Our main assumption is a coupling property which is expressed in terms of the one-step approximation. We show that the proposed set-up can be applied to a wide range of equations that may be law dependent, such as Langevin equations, reflected equations, Boltzmann type equations and for a recent McKean Vlasov type model for neuronal activity
Compétences et pratiques artistiques des illustrateurs à l'ère de l'intelligence artificielle générative d'images
No full textNational audienceL’ajout dans les années 2000, par Georgette Yakman, de l’art dans les programmes pédagogiques de type STEM (devenant ainsi STEAM) souhaitait mettre en avant la place centrale de l’émotion, de l’éthique ou encore de l’esprit critique dans l’inspiration et la création au sein d’un monde technologique (Calongne, 2024, p.2). L’arrivée récente des intelligences artificielles (IA) génératives d’images, comme Midjourney, DALL-E, Firefly ou encore Stable Diffusion, remet-t-elle en question cette vision du monde ? En effet, n’importe qui peut dorénavant produire des images sans aucune autre compétence que la rédaction d’un texte en pseudo-langage naturel (c’est-à-dire plutôt schématisé) appelé prompt ou modèle multimodal text-to-image (Messer, 2024). Dans ce contexte, de nombreux chercheurs et professionnels s’interrogent sur la capacité de création même de ces synthétiseurs d’images les appelant parfois « perroquets stochastiques » (Bender et al., 2021, p.616) tout en leur trouvant également une certaine marge de liberté (de production) dans la variation aléatoire de leurs processus de génération (Crevoisier, 2024). On ne compte d’ailleurs plus le nombre de textes ou d’émissions titrant, peu ou prou, l’IA générative comme une menace ou une opportunité pour la création… Dans ce débat, le monde de la culture est généralement appréhendé en deux groupes : l’IA Art d’une part, qui ne serait pas menacée car notamment soutenue par des fonds publics ou des mécènes privés, et les créateurs plus intégrés au tissu économique classique, voués à un questionnement quasi-existentiel, d’autre part. Toutefois, même dans le cadre de cette seconde approche, il nous semble qu’il serait fallacieux de réduire le métier d’illustrateur ou de graphiste à la seule production utilitariste d’images. Le travail de l’artiste va bien au-delà. Par conséquent, nous nous interrogeons dans ce chapitre sur l’évolution de quelques compétences et pratiques fondamentales des illustrateurs, et plus particulièrement des illustrateurs de jeux de société, à l’ère des IA génératives d’images
Decoding cortical folding with deep learning: toward neurodevelopmental biomarkers of psychiatric disorders
No full textInternational audienceCortical folding is primarily determined during prenatal and early postnatal brain development and remains relatively stable throughout life. Accumulating evidence suggests that psychiatric disorders, including schizophrenia (SCZ), Bipolar Disorder (BD), and Autism Spectrum Disorder (ASD), result from intricate interactions between early neurodevelopmental disruptions and subsequent environmental influences. Therefore, cortical folding patterns are promising candidates as stable imaging biomarkers reflecting the neurodevelopmental component of psychiatric conditions. This study aims to demonstrate that deep learning can be used to learn meaningful representations of cortical folding patterns that enable the individual-level prediction of major psychiatric disorders, namely SCZ, BD, and ASD. We introduce a dedicated deep learning architecture leveraging self-supervised pre-training on large datasets of the general population to extract subjectlevel representations of cortical folding from structural magnetic resonance imaging (MRI) data. We demonstrate that the learned representations allow significant single-subject prediction of clinical status for SCZ, BD, and ASD. The proposed folding-based representation presents a novel approach for identifying imaging biomarkers related to the neurodevelopmental origins of psychiatric disorders. It opens the possibility of inferring early-life brain alterations from adult MRI data, offering potential tools for stratification and precision psychiatry
Digestive and bariatric surgery, digestive endoscopy and interventional radiology
No full textInternational audienc
Advanced insights into the biodeterioration and conservation strategies of cultural heritage: A review
No full textInternational audienc
Organizing collective political action in a world in polycrisis: insights from Hannah Arendt and John Dewey
No full textInternational audienc
Evidence for the absence of a relationship between inflammation and cognition in a cohort of 1565 individuals with bipolar spectrum disorders: a Bayesian analysis of network
No full textInternational audiencePrevious studies have reported variable associations between peripheral inflammatory markers and cognitive functioning in individuals with bipolar spectrum disorders (BSD), with some identifying significant links and others finding no relationship. Such inconsistencies raise important questions about the role of inflammation in cognitive impairment among individuals with BSD. This study aims to investigate the relationship between peripheral inflammatory markers and cognitive function in a clinical sample of individuals with BSD using a Bayesian network analysis framework. We analyzed data from a large cohort (n = 1565) focusing on hsCRP and a subsample (n = 249) that included concurrent assessments of additional cytokines including Interleukin-6 and Tumor Necrosis Factor-alpha. A Bayesian approach was utilized to quantify uncertainty regarding the presence or absence of associations between inflammation and cognitive function. Our findings revealed no significant associations between inflammatory markers and cognitive performance in both samples. Strong evidence was found supporting the absence of association, with network analysis indicating distinct clusters for cognitive and inflammatory variables, suggesting they function as independent constructs with limited interactions. In our clinical sample of individuals with BSD, our findings do not support a direct association between some inflammatory markers and cognition, aligning with studies that found minimal or no associations. Our study emphasizes the importance of utilizing Bayesian methods to assess these relationships rigorously and suggests further exploration of individual differences and subgroup effects in future research