Portail HAL Nantes Université
Not a member yet
120169 research outputs found
Sort by
Médiatisation, dénonciation publique et prémices de l’indignation collective contre les violences sexuelles dans les années 1970 en France
No full textInternational audienceThis paper deals with the emergence of the violence against minors public denunciation in France by analyzing the press coverage for Leïla Sebbar’s essay entitled “We kill the little girls an investigation into mistreatment, abuse, murders, incest, rapes against minor girls under fifteen years old, from 1967 to 1977 in France” by (Stock, 1978). The analysis of 20 papers reveals various lexical fields mobilized by journalists to build a public denunciation. By combining a variety of lexical fields, recalling facts and presenting evidence, they lay the foundations for a collective indignation in the face of what minors suffer in terms of physical, sexual, incestual and psychological violence. In particular, the arguments developed by women's rights defenders and those who condemn patriarchy and criticize the established social order make it possible to identify the choices made by journalists and authors of the aforementioned papers. They allow further to qualify and nuance the positions taken by the editorial teams according to their nature.Nous nous intéressons aux prémices de la médiatisation en France des violences contre les mineures à travers l’analyse de la couverture réservée par la presse écrite à l’essai On tue les petites filles une enquête sur les mauvais traitements, sévices, meurtres, incestes, viols contre les filles mineures de moins de quinze ans, de 1967 à 1977 en France de Leïla Sebbar (Stock, 1978). L’analyse d’un corpus de 20 articles révèle les différents champs lexicaux mobilisés par les journalistes permettant de construire une dénonciation publique. La combinaison des différents champs lexicaux, le rappel des faits et la présentation d’éléments de preuve permettent de leur côté de jeter les bases d’une indignation collective face à ce que subissent les mineures comme violences physiques, sexuelles, incestuelles et psychologiques. En particulier, les argumentaires développés par les défenseurs des droits des femmes et ceux qui condamnent le patriarcat et critiquent l’ordre social établi permettent de cerner les choix opérés par les journalistes et les auteur.ices des articles ainsi que de nuancer les positionnements des rédactions en fonction de leur nature. Mots-clés : Médiatisation, dénonciation publique, indignation collective, violences sexuelles, presse écrite, années 1970
Formes de l’échec. Déclinaisons et enjeux d’une notion dans la tradition italienne et au-delà
No full textInternational audienc
Sphingobacterium spp. as intrinsic producers of SPI Ambler class B metallo-β-lactamases
No full textInternational audienceObjectives: To characterize the intrinsic carbapenemase of Sphingobacterium spp.Materials and methods: Sphingobacterium spp. clinical isolates (n = 63) were collected in several French laboratories from 2012 to 2022. Antibiotic susceptibility was assessed by disc diffusion. MICs of last resort antimicrobials (including carbapenems, ceftazidime-avibactam, cefepime-enmetazobactam and cefiderocol) were determined by broth microdilution. WGS was performed on all isolates using the Illumina method. Cloning of the Sphingobacterium spiritivorum β-lactamase encoding gene, blaSPI-1, allowed expression of a novel metallo-β-lactamase (MBL), SPI-1, in Escherichia coli. Enzymatic kinetic parameters were determined with β-lactams as substrates.Results: Among the 64 Sphingobacterium isolates collected we identified six species, of which S. spiritivorum (n = 24) was the most prevalent followed by S. multivorum (n = 20), S. siyangense (n = 16), S. lactis (n = 2), S. thalpophilum (n = 1) and S. faecium (n = 1). Sphingobacterium isolates displayed a phenotype of multidrug resistance including resistance to last resort β-lactams and β-lactam-β-lactamase inhibitors (with MIC50 at 4, 4 and 2 mg/L for ceftazidime-avibactam, cefepime-enmetazobactam and cefiderocol, respectively). The most effective drugs were fluoroquinolones, tigecycline and eravacycline. All S. spiritivorum displayed a carbapenemase activity that correlated with the identification of SPI-1 encoding genes. SPI-like variants (n = 10) shared 93.7% amino-acid identity, but only 50.6% identity with the closest CGB-1 MBL intrinsic of Chryseobacterium gleum. Characterization of SPI-1 showed a strong hydrolytic activity towards most β-lactams including carbapenems.Conclusion: S. spiritivorum was the most prevalent species of Sphingobacterium identified in clinical samples. All S. spiritivorum-like species produced a chromosomally encoded SPI-like MBL that might be challenging for the treatment of infections caused by this bacterial species
Can LLMs Revolutionize the Design of Explainable and Efficient TinyML Models?
No full textNational audienceThis paper introduces a novel framework for designing efficient neural network architectures specifically tailored to tiny machine learning (TinyML) platforms. By leveraging large language models (LLMs) for neural architecture search (NAS), a vision transformer (ViT)-based knowledge distillation (KD) strategy, and an explainability module, the approach strikes an optimal balance between accuracy, computational efficiency, and memory usage. The LLM-guided search explores a hierarchical search space, refining candidate architectures through Pareto optimization based on accuracy, multiply-accumulate operations (MACs), and memory metrics. The best-performing architectures are further fine-tuned using logits-based KD with a pre-trained ViT-B/16 model, which enhances generalization without increasing model size. Evaluated on the CIFAR-100 dataset and deployed on an STM32H7 microcontroller (MCU), the three proposed models, LMaNet-Elite, LMaNet-Core, and QwNet-Core, achieve accuracy scores of 74.50%, 74.20% and 73.00%, respectively. All three models surpass current state-of-the-art (SOTA) models, such as MCUNet-in3/in4 (69.62% / 72.86%) and XiNet (72.27%), while maintaining a low computational cost of less than 100 million MACs and adhering to the stringent 320 KB static random-access memory (SRAM) constraint. These results demonstrate the efficiency and performance of the proposed framework for TinyML platforms, underscoring the potential of combining LLM-driven search, Pareto optimization, KD, and explainability to develop accurate, efficient, and interpretable models. This approach opens new possibilities in NAS, enabling the design of efficient architectures specifically suited for TinyML. To facilitate further research and development in this field, the proposed framework and the best-performing architectures are made publicly available at Link
MUVOD: A Novel Multi-View Video Object Segmentation Dataset and a Benchmark for 3D Segmentation
No full textNational audienceThe application of methods based on Neural Radiance Fields (NeRF) and 3D Gaussian Splatting (3D GS) have steadily gained popularity in the field of 3D object segmentation in static scenes. These approaches demonstrate efficacy in a range of 3D scene understanding and editing tasks. Nevertheless, the 4D object segmentation of dynamic scenes remains an underexplored field due to the absence of a sufficiently extensive and accurately labelled multi-view video dataset. In this paper, we present MUVOD, a new multi-view video dataset for training and evaluating object segmentation in reconstructed real-world scenarios. The 17 selected scenes, describing various indoor or outdoor activities, are collected from different sources of datasets originating from various types of camera rigs. Each scene contains a minimum of 9 views and a maximum of 46 views. We provide 7830 RGB images (30 frames per video) with their corresponding segmentation mask in 4D motion, meaning that any object of interest in the scene could be tracked across temporal frames of a given view or across different views belonging to the same camera rig. This dataset, which contains 459 instances of 73 categories, is intended as a basic benchmark for the evaluation of multi-view video segmentation methods. We also present an evaluation metric and a baseline segmentation approach to encourage and evaluate progress in this evolving field. Additionally, we propose a new benchmark for 3D object segmentation task with a subset of annotated multi-view images selected from our MUVOD dataset. This subset contains 50 objects of different conditions in different scenarios, providing a more comprehensive analysis of state-of-the-art 3D object segmentation methods. Our proposed MUVOD dataset is available at https://volumetric-repository.labs.b-com.com/#/muvod
L’impact des patients enseignants sur l’approche centrée patient des internes en médecine générale : étude dans les Groupes d’Échange de Pratiques à Nantes - Session Patients impliqués
No full textInternational audiencePLANIFICATIONL’objectif de ce travail est d’identifier si l’intégration des patients enseignants (PE) dans les Groupes d’Échange de Pratiques (GEP) modifie l’approche centrée patient des Internes en médecine générale (IMG) à Nantes.INTERVENTIONÀ Nantes, les patients enseignants (PE) ont intégré les GEP au cours de l'année 2023-2024. Parmi les 33 groupes organisés pour les étudiants de la première à la troisième année d'internat, 10 ont bénéficié de la participation d'un PE.Une étude quantitative a été menée, en auto-évaluation par les IMG (échelle PPOS), ainsi qu'en hétéro-évaluation parmi les PE et médecins enseignants (échelle d’humanisme). Les questionnaires ont été soumis lors de la dernière séance à 10 groupes avec PE et 10 groupes sans PE.Les analyses ont été faites en tenant compte du sexe, de l'année d'étude, et de la présence ou non de PE.Les perceptions des participants sur le dispositif ont été recueillies de manière qualitative.ÉVALUATIONLes étudiants bénéficiant de PE dans leurs GEP, toutes années confondues, s'évaluent avec des scores PPOS plus faibles et des scores d’humanisme plus élevés, traduisant une orientation plus centrée patient.Le score d’humanisme donné par PE et médecins enseignants aux étudiants bénéficiant de PE baisse au cours de l’internat, traduisant une orientation plus centrée médecin ; cela pose aussi la question des attentes des enseignants envers les étudiants.Les retours qualitatifs soulignent un impact positif sur la communication et l'empathie des étudiants, mais mettent en évidence des freins structurels, notamment une organisation des GEP parfois inadaptée et une liberté d'expression restreinte.L'intégration des PE dans les GEP semble favoriser l'apprentissage de l'ACP, avec des bénéfices visibles en termes d'humanisme. Cependant, des limites freinent leur pleine intégration, et les écarts entre auto-évaluations et hétéro-évaluations interrogent une subjectivité dans l'évaluation de l'ACP.Les PE semblent renforcer l'approche centrée patient dans la formation médicale, à condition de repenser les dispositifs pédagogiques pour maximiser leur impact
Best practices guidelines for addressing social innovation into coastal rewilding
No full textThese best practices guidelines have been developed and designed to inform and support both local and global Multi-Actor Laboratories (MALs), as stakeholder engagement processes within REWRITE (See also Stakeholder Engagement Plan), by showcasing exemplary approaches to social innovation in coastal rewilding. Based on a set of criteria, a substantial number of sites across Europe were selected as emblematic experiences and systematically investigated to extract lessons learned and identify specific forms of social innovation and stakeholder engagement. The resulting material contributes to the guidelines in two complementary ways: (1) by providing a comprehensive overview of the levels and methods of stakeholder participation employed in coastal rewilding initiatives, and (2) by identifying representative cases that are presented as detailed Case Studies. A total of 19 Case Studies are highlighted, with at least one case study for each thematic focus (Table 1). Following a concise overview of coastal rewilding practices and definitions, as well as an overview of past and ongoing projects implemented in Western Europe, the core structure of the guidelines is organized around three main entry points.First, the guidelines explore the spectrum of stakeholder engagement in decision-making, drawing on a simplified adaptation of Arnstein’s (1969) ladder of participation. This framework ranges from basic information dissemination and communication, through public consultation and stakeholder involvement, to stakeholder collaboration and, ultimately, stakeholder empowerment.Second, a dedicated section addresses participatory coastal rewilding, outlining four stages of the process: building a collective diagnostic, co-designing visions and co-planning rewilding scenarios, co-implementing projects and practices, and co-monitoring to guide adaptive management. An additional category is included to describe mechanisms that enable more permanent participatory processes, ensuring the continuity of rewilding initiatives over time.Third, the guidelines highlight the communication and valorization of the benefits of coastal rewilding, recognizing that effective messaging can enhance stakeholder engagement and incentivize participation. Five specific dimensions are examined in this context: reinforcing traditional cultures and local economies, promoting nature-based tourism, delivering educational programs, providing stakeholder training, and leveraging narratives and storytelling to foster broader understanding and support.Together, these three entry points provide a structured and practical framework for understanding, implementing, and sustaining socially inclusive coastal rewilding practices
Suivi pédiatrique de l’enfant de 3 à 6 ans : conception de modules d’enseignement distanciel pour les internes en médecine générale - Session Suivi de l'enfant
No full textInternational audienceCONTEXTEFace à la diminution du nombre de pédiatres en France, les médecins généralistes assurent désormais une part croissante des consultations pédiatriques. Dans ce contexte, la formation en pédiatrie des internes en médecine générale constitue un enjeu majeur. Pourtant, et malgré la réalisation d’un stage de pédiatrie au cours de leur DES, cet enseignement reste hétérogène selon les territoires et souvent jugé insuffisant par les intéressés eux-mêmes.L'objectif de ce travail est la conception de quatre modules d’enseignement distanciel dédiés au suivi, au dépistage, à la prévention et à l’orientation des enfants âgés de 3 à 6 ans, intégrés à un parcours pédagogique regroupant un corpus d’enseignements hybrides en pédiatrie.PLANIFICATIONUne démarche inspirée du guide méthodologique de l’Université Laval et échelonnée sur deux ans a structuré la création des modules : recherche documentaire, conception pédagogique, production de contenus, relecture, diffusion puis évaluation.Le projet s’est appuyé sur une collaboration étroite entre médecins généralistes, pédiatres, acteurs de santé publique régionaux et ingénieurs pédagogiques, avec le soutien de l’université.INTERVENTIONQuatre modules de 2 à 3 heures ont été réalisés, comprenant chacun une base théorique vocalisée, des activités pédagogiques interactives, des ressources complémentaires et une fiche résumé téléchargeable.Les thématiques sont :Suivi de l’enfant de 3 à 6 ans (développement, prévention et examens systématiques)Environnement de l’enfant (polluants, perturbateurs endocriniens, accidents domestiques)Surpoids et obésité de l'enfantSommeil (physiologie, troubles, conseils parentaux)ÉVALUATIONCe projet constitue une innovation pédagogique adaptée aux contraintes des internes. L’enseignement distanciel favorise l’accessibilité et l’autonomie des apprenants. Néanmoins, il présente des limites, notamment l'absence d'interaction directe avec les formateurs et de mise en situation pratique.Une évaluation formative est prévue afin de mesurer l’évolution du sentiment d’efficacité personnelle des apprenants, d'adapter et de développer l'outil pour répondre au mieux aux besoins de formation des étudiants.Ces modules distanciels illustrent une réponse structurée et adaptable aux besoins croissants de formation pédiatrique en soins primaires. Leur diffusion pourrait être élargie à d’autres Départements de Médecine Générale, favorisant une harmonisation des apprentissages à l’échelle nationale
Synthesis, Photophysical properties and Ab Initio Calculations of Dual Solution-Solid-State Emitting Ethynyl-Extended Boranil Complexes
No full textInternational audienceWe describe herein the synthesis along with full photophysical and computational studies of three series of boranil complexes (Series I-III), incorporating extended ethynyl-substitution, either on the imino side (Series I), on the phenolic side (Series II), or on both sides (Series III), in order to assess the impact of the insertion of electron donors or acceptors on the fluorescence properties. A full photophysical study in solution (various solvents) demonstrated the presence of a strong intramolecular charge transfer process within these boron complexes, with fluorescence colors spanning the whole visible range. Additionally, these compounds are also emissive in the solid-state, on a similar wide fluorescence range thus providing more examples of dual solution-and solid-emissive fluorophores, based on boron complexes. Theoretical calculations on both anil ligands and boranil complexes rationalize the charge transfer nature of the excited-states involved in the emissive transitions
Inclusive interaction at a virtual scale: using multi-user virtual environments in the educational workflow of architectural design
No full textInternational audienceThis study introduces the integration of multi-user virtual environments (MUVEs) in a proposed four-phased workflow for architecture education. The workflow responds to the growing reports of students’ difficulties and dissatisfaction with interaction on online platforms. MUVEs are integrated as computational means to enhance inclusion in the learning process. Inclusion is reflected in active participation and interaction between culturally diverse participants. Free from real-world constraints, MUVEs allow learners to reconsider artefacts’ scale as stimuli for interaction. The study describes how MUVEs are used and experienced in a workshop given simultaneously to architecture students from France and India. Observations, outcome analysis and a post-course questionnaire were used to assess learning activity and inclusive interaction. Findings exhibit differences in accomplishing the workflow phases and utilising scale. The study identifies four unique concepts using MUVEs to facilitate inclusive interaction. This research brings new insights into how MUVEs can be integrated to foster quality education