HAL - La Rochelle Université, Archives Ouvertes
Not a member yet
19516 research outputs found
Sort by
AutoDFL: A Scalable and Automated Reputation-Aware Decentralized Federated Learning
No full textInternational audienceBlockchained federated learning (BFL) combines the concepts of federated learning and blockchain technology to enhance privacy, security, and transparency in collaborative machine learning models. However, implementing BFL frameworks poses challenges in terms of scalability and cost-effectiveness. Reputation-aware BFL poses even more challenges, as blockchain validators are tasked with processing federated learning transactions along with the transactions that evaluate FL tasks and aggregate reputations. This leads to faster blockchain congestion and performance degradation. To improve BFL efficiency while increasing scalability and reducing on-chain reputation management costs, this paper proposes AutoDFL, a scalable and automated reputation-aware decentralized federated learning framework. AutoDFL leverages zk-Rollups as a Layer-2 scaling solution to boost the performance while maintaining the same level of security as the underlying Layer-1 blockchain. Moreover, AutoDFL introduces an automated and fair reputation model designed to incentivize federated learning actors. We develop a proof of concept for our framework for an accurate evaluation. Tested with various custom workloads, AutoDFL reaches an average throughput of over 3000 TPS with a gas reduction of up to 20X
Circumpolar spread of avian influenza H5N1 to southern Indian Ocean islands
No full textVoir aussi l'actualité CNRS INEE du 26/01/2026 : L’Influenza Aviaire Hautement Pathogène menace la biodiversité subantarctique dans le sud de l’Océan Indien : https://www.inee.cnrs.fr/fr/cnrsinfo/linfluenza-aviaire-hautement-pathogene-menace-la-biodiversite-subantarctique-dans-le-sudInternational audienceSince 2020, the outbreak of high pathogenicity avian influenza (HPAI) H5N1 virus clade 2.3.4.4b has turned into the largest documented panzootic 1,3 . Here, we describe its arrival into the Indian Ocean sub-Antarctic archipelagos of Crozet and Kerguelen, where we first detected the virus in October 2024 in dead southern elephant seals. While the panzootic is ongoing, it has already caused unprecedented mortalities of marine mammals and seabirds. We collected brain swabs from seal and seabird carcasses and obtained 25 novel HPAI H5N1 2.3.4.4b sequences. Using phylogeographic analyses, we show that there have been independent introductions of the virus to Crozet and Kerguelen islands, most likely from the distant South Georgia islands in the Southern Atlantic, and not from the more nearby coasts of South Africa. Our results point to a year-long gap in genomic surveillance in the sub-Antarctic region. Locally, our analyses show that the virus is transmitted between different species. Our serological analyses show that some southern elephant seal had mounted an anti-H5 antibody response. Through its circumpolar spread to the Indian Ocean, HPAI H5N1 2.3.4.4b moves closer to Australia, which remains free from infections with this strain, and represents a major threat to the sub-Antarctic wildlife.The ongoing outbreak of clade 2.3.4.4b of high pathogenicity avian influenza (HPAI) H5N1 virus has been unique not only in the large number of host species affected, with a wide range of both birds and mammals reported dead 1-3 , but also in the scale of the panzootic, with all continents except Oceania affected 4 . Originating in Asia in 1996 5,6 , the A/Goose/Guangdong/1/1996 lineage HPAIV spread across the world in different waves, recombined and diversified into different clades of various H5Nx viruses. Clade 2.3.4.4b started to become a worldwide preoccupation in 2014, and in the large outbreak of 2016-2017, HPAI H5N8 viruses of clade 2.3.4.4b reached South Africa 7,8 . Since 2021, HPAI H5N1 viruses of clade 2.3.4.4b severely impacted wild bird species and poultry across Europe 9-11 , southern Africa 12-14 , North America 15 , and South America where it resulted in die-offs of tens of thousands of seabirds 16 and marine mammals 17,18 . At the time of writing, HPAIV H5N1 clade 2.3.4.4b has not been detected in any birds or other animals in Australia and the southern Pacific region 19 . While there were no incursions into the Antarctic region during the summer season 2022/2023 20 , HPAIV H5N1 clade 2.3.4.4b was detected for the first time</div
Quel avenir pour les causes d'irresponsabilité pénale ? Propos introductifs
No full textInternational audienc
HIBOU 2030: an integrated method for the Hybrid assessment of the Interactions between the BiOdiversity, the nature-based solutions, and the Urban system
No full textInternational audienceThe objective of this paper is to present the integrated method HIBOU 2030, which is employed to assess the efficiency of the Nature-Based Solutions (NBS) and their alternatives (e.g. grey, hybrid solutions) for urban projects. The HIBOU 2030 method aligns with international initiatives, such as the Science-based targets Network for nature[1] that promote the integrated assessment approaches. The HIBOU 2030 method is thus design to place the urban system integrating the NBS and theirs alternatives at the core of its approach, with the objective of contributing to several of the action-oriented global targets for 2030 outlined in the Global Biodiversity Framework (GBF)[2] such as: Target 11 - Restore, Maintain and Enhance Nature’s Contributions to People, the Target 12 - Enhance Green Spaces and Urban Planning for Human Well-Being and Biodiversity and the Target 14 - Integrate Biodiversity in Decision-Making at Every Level. HIBOU 2030 is based on the hybridization of several area of expertise (e.g. Life Cycle Assessment, ecology, urban planning, etc.) and its fundamental principles are as follows: 1) interactions (both positive and negative ) between biodiversity, NBS and the urban project occur on the project site (in situ) but also on global scale (ex-situ) 2) the multifunctionality of NBS is one of the answers to numerous urban challenges and it must be taken into account in the analysis of the results; 3) the integrated approach necessitates the establishment of a shared semantics among the various fields of expertise; a common macro-model to characterize the system to be assessed and the different development options; the interdependence of results for each issue. Consequently, a parameter variation to address one of the questions will inherently influence the others.HIBOU 2030 method and its associated toolset facilitate the assessment the urban project’s contribution to the following urban challenges: 1) the climate change (1 indicator); 2) the biodiversity in situ and ex situ (8 indicators). These indicators are designed to address as much as possible of the five pressures on the biodiversity: global warming, land use change, pollution, overexploitation of resources, introduction of invasive species; 3) the stormwater management (1 indicator); 4) the urban heat island (UHI) mitigation (1 to 3 indicators); 5) the urban quality for the citizens, based on a qualitative assessment grid considering 24 criteria.HIBOU 2030 is a tool used to conduct expertise and research studies, thereby supporting various stakeholders’ analyses and decision-making processes concerning construction and renovation actions for buildings and urban projects. Continuous improvement is achieved through the collection and analysis of feedback from its use in various European contexts. [1]https://sciencebasedtargetsnetwork.org [2]https://www.cbd.int/gbf/target
Lightweight Visualisation for Vortex Tracking in Airflow Acquisition
No full textInternational audienceVentilation systems are spread in most buildings and housing. They provide control on air quality, while provisioning acceptable thermal conditions. However, slotted plates that direct air jets often produce acoustic disturbances with a self-sustained tone. Understanding and controlling the phenomenon requires complex experimentation with expensive setups. This article proposes an open source, web-based, interactive visualisation system dedicated to the observation and analysis of recorded high frequency captures (3kHz or more) of air streams, charged with thin oil particles. It relies on the acquired images and estimated vector fields, coming from industrial existing systems. Our goal is to visualise the main flow parameters, such as speed, gradients, directions, vortex tracking with path prediction, and vortex frequencies. The obtained results highlight interesting phenomena that illustrate sounds production. They can be employed by physicists to understand, explain, and control the generated acoustics
eDNA and Citizen Science Reveal Hidden Fish Biodiversity in Climate‐Stressed Urban Ports of the Mediterranean Sea
No full textInternational audienceThis paper provides a pioneering study case on monitoring fish biodiversity in ports through the eDNA and citizen science approach. eDNA samples were collected in the spring and in fall 2022 in the ports of Calvi,</div
Le prestre et Alison - Édité d'après le ms. BnF fr. 19152
No full textLe prestre et Alison - Édité d'après le ms. BnF fr. 1915
Le roman de un chivaler e de sa dame e de un clerk - Édité d'après le ms. Cambridge, Corpus Christi Coll., 50
No full textLe roman de un chivaler e de sa dame e de un clerk - Édité d'après le ms. Cambridge, Corpus Christi Coll., 5
La bataille des vins d'Henri d'Andeli - Édité et traduit d'après le ms. BnF fr. 837
No full textLa bataille des vins - Henri d'Andeli - édité et traduit d'après le ms. BnF fr. 83
Agricultural intensification degrades arable weed diversity into phylogenetically unstructured subsets of tolerant species
No full textInternational audienceContext: The rise of agricultural intensification (AgI) has severely impacted arable weeds, making it crucial to understand how this process shapes their assemblages across agricultural landscapes.Objectives: To elucidate how variation in species composition (beta diversity) among arable weed assemblages respond to AgI gradients, with a particular focus on whether phylogenetic relationships structure these patterns.Methods: We analyzed farm-level arable weed assemblages across nine European regions with distinct agricultural management contexts, focusing on in-field AgI indicators (yield, nitrogen fertilizer, pesticides) and landscape context (field size, cultivated area, crop diversity). To examine compositional changes among assemblages, we partitioned beta diversity into its turnover and nestedness components.Results: We found positive correlations between in-field AgI differences among farms and the nestedness component of taxonomic beta diversity, alongside a previously reported negative relationship between in-field AgI and species richness. In contrast, the landscape dimension of AgI had a comparatively minor effect. Phylogenetic structure metrics showed weak and inconsistent responses to AgI.Conclusions: In-field variation in AgI—rather than landscape context—contributes significantly to taxonomic dissimilarity among arable weed assemblages on European farms, with increasing AgI driving the exclusion of sensitive species and the persistence of tolerant ones, without evidence of species potentially restricted to highly intensified conditions. Traits associated with AgI likely evolved through distinct and complex evolutionary trajectories long before the surge of AgI in the mid-twentieth century, which may explain the phylogenetically unstructured patterns observed