HAL Université de Savoie
Not a member yet
59053 research outputs found
Sort by
EFFECTS OF CLASS DISTRIBUTION SHIFT IN MULTI-LABEL CLASSIFICATION SPLITS: A STUDY CASE ON BIGEARTHNET
No full textInternational audienceClass imbalance is a common issue for multi-label classification tasks, and often induces class distribution shift between training, validation and test splits. Land-use land-cover classification tasks in remote sensing are, inherently, often multilabel problems. Training deep learning models on benchmark datasets that address this issue is then a key component to ensure high performance. This paper proposes the application of an iterative stratified sampling method to rebalance splits of a benchmark dataset. The BigEarthNet dataset [1] is used as a study case to show the improved training performances of convolutional neural networks trained on the resampled data compared to those trained on the original randomly split data
L'urbanisation et l'intensité de gestion des sols urbains modifient le cycle de l'azote et les fonctions microbiennes
No full textInternational audienceL’urbanisation modifie en profondeur les cycles biogéochimiques du sol, notamment celui de l’azote (N), à travers l’usage des sols, l’intensité de leur gestion et les apports exogènes. Ces transformations affectent les communautés microbiennes et le fonctionnement du sol, ce qui peut réduire la fertilité, entraîner des pertes de nutriments et accroître les risques environnementaux (pollution des eaux, émissions de gaz à effet de serre). Cette étude explore comment différents types d’usages des sols urbains – jardins vitrine, parcs, zones arborées, bords de routes, zones résidentielles, terrains de sport et réserves foncières – influencent les processus du cycle de l’azote et la structure fonctionnelle des communautés microbiennes. L’étude a combiné des approches biogéochimiques (concentrations de NH₄ ⁺, NO₃ ⁻, azote total), fonctionnelles (taux de minéralisation, nitrification et dénitrification potentielle) et moléculaires (abondance des gènes fonctionnels liés au cycle de l’azote, tels que amoA, nirK, nirS, nosZI, nosZII) couplées à des analyses de diversité (séquençage Illumina MiSeq) pour caractériser les fonctions microbiennes impliquées dans la transformation de l’azote. Les résultats montrent une forte variabilité de la dynamique du cycle de l’azote selon l’usage du sol. Les jardins vitrine urbains présentent des concentrations en nitrate significativement plus élevées, attribuées à une fertilisation minérale intense, à une nitrification accrue et à une abondance plus élevée de gènes amoA. À l’inverse, le potentiel de dénitrification est plus marqué dans les parcs, zones résidentielles et bords de route, où les sols sont plus humides et moins aérés ; conditions favorables aux bactéries dénitrifiantes. La modélisation par équations structurelles (SEM) a permis d’identifier les facteurs clés du fonctionnement du cycle de l’azote en milieu urbain. Le carbone organique du sol (SOC) apparaît comme un déterminant central de la dénitrification, tandis que le pH est fortement corrélé à l’abondance des archées oxydatrices de l’ammonium (AOA). Sur le plan de la composition des communautés bactériennes, l’étude révèle que les fonctions liées au cycle de l’azote varient selon l’usage du sol. Les zones très peu gérées (réserves foncières) abritent davantage de bactéries fixatrices d’azote, tandis que les zones fortement gérées (jardins vitrine, terrains de sport) montrent une prédominance de bactéries impliquées dans la nitrification. Par ailleurs, l’absence de corrélation significative entre les différentes étapes du cycle (minéralisation, nitrification, dénitrification) dans les zones gérées intensivement indique une rupture du cycle de l’azote, probablement liée à l’aération excessive, au travail du sol et à la faible accumulation de matière organique. Enfin, des différences notables de composition des communautés végétales et de gestion influencent également la dynamique de l’azote. Ces résultats soulignent la sensibilité du cycle de l’azote à l’usage des sols en milieu urbain et offre des leviers pour une gestion plus durable de ces milieux : limitation de la fertilisation minérale, conservation de la matière organique, réduction du compactage et mise en place d’une végétation plus diversifiée. Elle plaide pour une meilleure intégration des processus biogéochimiques dans la planification urbaine, afin de renforcer la résilience des écosystèmes urbains face aux pressions anthropiques et climatique
Could invisible effects of land use and management on soil microbial communities durably impact carbon and nitrogen cycling in urban soils?
No full textInternational audienceUrbanization profoundly alters nitrogen and carbon cycling through changes in land use and management intensity, affecting both soil functions and microbial communities. These shifts can degrade soil fertility and disrupt ecosystem processes. This study investigates how different urban land use types—including showcase gardens, parks, tree-covered areas, roadsides, residential zones, sport fields and unused areas—influence nitrogen dynamics and microbial functions. We used a multifaceted approach combining biogeochemical measurements (NH₄⁺, NO₃⁻, total N and C), functional assays (mineralisation, nitrification, potential denitrification), organic matter characterization (via Rock-Eval analysis), and molecular techniques (functional gene abundance and bacterial community profiling using Illumina MiSeq).Results show land use significantly influences nitrogen processes. Showcase garden areas showed elevated nitrate levels, driven by fertilization and higher nitrification activity, whereas denitrification potential and related gene abundances were higher in parks and roadside soils. Structural equation modelling (SEM) revealed that soil organic carbon is a major driver of denitrification, and pH influences the abundance of ammonia-oxidizing archaea (AOA). Water retention, often reduced in compacted soils, negatively affected the nitrogen cycle.In terms of bacterial communities, the study reveals that bacterial functions linked to the nitrogen cycle vary according to land use. Lightly managed areas harbour more nitrogen-fixing bacteria, while intensely managed areas show a predominance of bacteria involved in nitrification. Furthermore, the absence of any significant correlation between the different stages of the cycle (mineralisation, nitrification, denitrification) in intensively managed areas indicates a breakdown in the nitrogen cycle, probably linked to excessive aeration, tillage and low organic matter accumulation or to difference in organics matter decomposition and stability.These findings highlight the sensitivity of nitrogen cycling to urban land management. Crucially, the role of organic matter dynamics in this context requires deeper investigation. Future analysis using Rock-Eval will be key to understanding the quality, stability, and thermal reactivity of soil organic matter and its influence on microbial nitrogen transformations. This approach can reveal how different land uses affect organic matter decomposition and its capacity to support sustained microbial activity
Metabolic resistance of the tiger mosquito to pyrethroid insecticides in La Réunion Island likely results from local adaptation
No full textAbstract The resistance of mosquitoes to insecticides is a valuable model system for studying the genetic bases of xenobiotic adaptation in insects. The spread of the Asian tiger mosquito Aedes albopictus combined to the massive use of pyrethroid insecticides to limit arbovirus transmission resulted in the rise of resistance in various continents. Here, we investigated the genetic mechanisms underlying the recent adaptation of this mosquito to deltamethrin in La Réunion island. Bioassays confirmed the presence of resistance alleles in field populations. The resistance phenotype was further enhanced in the laboratory following a few generations of controlled selection. Combining whole genome Pool-seq and RNA-seq revealed no evidence of target-site resistance mutations but the over-expression and variant selection of detoxification enzymes associated with pyrethroid metabolism including cytochrome P450s, transferases and ABC-transporters. Among over-expressed detoxification genes, only one was linked to a gene duplication while polymorphism data suggest most of them being trans-regulated. Genome-wide selection signatures revealed a 9 Mb inverted superlocus responding to insecticide selection whose phenotypical importance remains uncertain. Altogether, this study indicates that the multigenic metabolic resistance phenotype observed in this insular territory mainly results from local adaptation. From an applied perspective, this study provides a set of markers to track pyrethroid resistance in the tiger mosquito in the South-West Indian Ocean. As this region is subjected to recurrent arbovirus outbreaks, the additive resistance phenotype that may arise from the introduction of Kdr mutations from other territories also calls for improving resistance surveillance at the regional scale. Author summary While novel vector control strategies are being developed, chemical insecticides remain widely used to control mosquitoes transmitting human diseases such as the Asian tiger mosquito. However, the recurrent use of insecticides resulted in the emergence of resistance which can ultimately affect vector control efficacy. Here, we investigate the genetic bases underlying the resistance of the Asian tiger mosquito to the pyrethroid insecticide deltamethrin in La Réunion island. By combining two complementary genomic approaches, we showed that resistance is mainly caused by an increased insecticide detoxification while classical ‘Knock down resistance’ mutations affecting the target of the insecticide were not detected. We also showed that resistance is underlain by multiple genetic changes spread across the genome, supporting the local selection of resistance rather than the introduction of resistance alleles. Furthermore, we identified a large inverted supergene responding to insecticide selection. This study provides valuable insights into the genetic bases of insecticide resistance, enabling the implementation of molecular makers to improve the tracking of insecticide resistance in this major mosquito vector across the Indian Ocean
Comparative study of characterization techniques in the Millimeter Waves and THz domains
No full textInternational audienceThe Millimeter Wave (mmW) and Terahertz (THz) domains are very dynamic fields in term of applied research, particularly with a view to using them for high-frequency telecommunications [1]. In this context, the characterization of new ranges of materials is becoming a topical issue. THz Time Domain Spectroscopy (THz- TDS) technique [2,3] is commonly used for precise characterization of materials in a frequency range from 100 GHz to around 5 THz. On the other hand, with increasing frequency of radiofrequency (RF) characterization tools, such as FSM-CW (Free Space Method with Continuous Wave), new characterization methods are now available to probe the lower part of this spectral range, up to several hundreds of GHz [4,5]. In this work, we compare the two main techniques, THz-TDS and FSM-CW, in terms of characteristics and performance (bandwidths, dynamics, limits, uncertainties, reproducibility…). The THz-TDS is a time-domain method that is intrinsically very broadband as the whole spectra is reached with only one measurement. The FSM-CW method is a frequency-domain approach that utilizes a Vector Network Analyzer (VNA). Thus, to span a frequency range as broad as that covered by the THz-TDS method, multiple setups are required. In this study we probed the following bands: W (75–110 GHz) and J (220–330 GHz). By using these different techniques and setups, we characterized a set of different materials to cover a wide range of permittivity values, for which we also varied the thickness. Figure 1 exhibits results we obtained on HDPE, PA6 and Al2O3 with 3 different thicknesses (uncertainties on the real part of the permittivity have not been plotted for sake of clarity). Due to the low value of the imaginary part of the permittivity, the extraction process sometimes gives negative values, that must be considered in regards to all the uncertainties. All the results will be presented and commented at the conference. This work is supported by the France 2030 programs, grant ANR-22-PEFT-0006 (NF-SYSTERA, PEPR 5G and beyond - Future Networks)
Probing Standard Model-like di-Higgs Production at Photon-Photon Colliders in the I(1+2)HDM Type-I
No full textInternational audienceIn this paper, pair production of Standard Model (SM)-like Higgs bosons, , is studied through scattering at future electron-positron colliders, in the framework of the Inert Doublet Model with two Active Doublets, i.e., the I(1+2)HDM for short. The relevance of the process for such a Beyond the SM (BSM) scenario stems from the fact that it is a one-loop process at lowest order, wherein inert charged states contribute alongside with , and heavy fermions (primarily, bottom and top quarks), crucially, at the same perturbative order. {Given that masses and () couplings are very mildly constrained,} there exist regions of the parameter space of the I(1+2)HDM where the former can be rather light and the latter rather large. After imposing up-to-date theoretical and experimental constraints on the I(1+2)HDM, it is found that the production rates of such process at future machines can be enhanced up to a factor of with respect to the SM, significantly exceeding typical yields of conventional 2-Higgs Doublet Models (2HDMs). Further, thanks to the level of control that one can attain at such facilities on the photon kinematics, leading to excellent invariant mass resolution of the incoming photon pairs, we show how it is possible to extract from this process the value of the mass (along that of the active states) with high precision, whichever the decays of the pair, both with and without beam polarization
Designing Metastable P3‐type Layered Negative Electrodes with High Na Vacancy Concentration for High‐Power Sodium‐Ion Batteries
No full textInternational audienceAbstract Titanium‐based layered oxides exhibit promising characteristics for negative electrode applications for sodium‐ion batteries (SIBs) owing to their low operating potential and stable redox behavior. However, challenges persist in synthesizing single‐phase Na x Cr x Ti 1– x O 2 materials with reduced sodium content ( x < 0.58) due to structural instability. In this study, an unconventional approach utilizing potassium analogues to design high Na vacancy concentration compounds is proposed. By exploiting the larger ionic size of K + ions, a P3‐type K 0.5 Cr 0.5 Ti 0.5 O 2 layered material with lower alkali ion concentrations ( x = 0.50) is stabilized. Subsequently, through a facile room‐temperature K + /Na + ion‐exchange process, sodium‐deficient metastable P3‐type Na 0.5 Cr 0.5 Ti 0.5 O 2 is successfully synthesized. A similar K + /Li + ion‐exchange process is also applied to synthesize O3‐type Li 0.5 Cr 0.5 Ti 0.5 O 2 . X‐ray diffraction combined with electron microscopy reveals the formation of metastable single‐phase P3‐type Na 0.5 Cr 0.5 Ti 0.5 O 2 and metastable O3‐type Li 0.5 Cr 0.5 Ti 0.5 O 2 with a unique layer arrangement. The high Na vacancy concentration in P3‐type Na 0.5 Cr 0.5 Ti 0.5 O 2 results in an increased initial capacity of 125 mA h g −1 at 10 mA g −1 . Additionally, Na 0.5 Cr 0.5 Ti 0.5 O 2 exhibits Na + /vacancy disordering and high electronic conductivity, enabling a high‐rate charge capability without sodium plating. This work provides new insights into the design of metastable layered materials for durable and safe sodium battery applications with fast‐charge capability
The track-length extension fitting algorithm for energy measurement of interacting particles in liquid argon TPCs and its performance with ProtoDUNE-SP data
No full textInternational audienceThis paper introduces the hypothetical track-length fitting algorithm, a novel method for measuring the kinetic energies of ionizing particles in liquid argon time projection chambers (LArTPCs). The algorithm finds the most probable offset in track length for a track-like object by comparing the measured ionization density as a function of position with a theoretical prediction of the energy loss as a function of the energy, including models of electron recombination and detector response. The algorithm can be used to measure the energies of particles that interact before they stop, such as charged pions that are absorbed by argon nuclei. The algorithm's energy measurement resolutions and fractional biases are presented as functions of particle kinetic energy and number of track hits using samples of stopping secondary charged pions in data collected by the ProtoDUNE-SP detector, and also in a detailed simulation. Additional studies describe impact of the dE/dx model on energy measurement performance. The method described in this paper to characterize the energy measurement performance can be repeated in any LArTPC experiment using stopping secondary charged pions
Envisioning Abolition
No full textInternational audienceAbolitionist thought visualises a world without prisons – or a radical reduction or transformation of prisons and punishment. This fascinating book explores the abolitionist ideas of key early socialists and anarchists, writing from the late eighteenth to the early nineteenth centuries. It considers how these radical thinkers can provide insights into our present condition, both by highlighting the harms of punishment and by pointing to inspiring alternatives to current policy and practice.By examining their calls for the ending of legal coercion, domination and repression, the book shows how the ideas of early socialists and anarchists can assist those engaging in emancipatory struggles against penal and social injustice today
Fibre Optic-Based Patch Sensor for Crack Monitoring in Concrete Structures
No full textInternational audienceIn the context of the necessity to monitor civil engineering structures for accident prevention, this paper presents the performance of an innovative optical fibre patch sensor that can be embedded or glued on the concrete's surface. The sensor is composed of a fabric comprising an optical fibre network. When employed in concrete structures, it is capable of detecting opening cracks in the order of 0.01 mm, thereby satisfying the regulatory standards for limit states (Eurocode2). This sensor is capable of accurately tracking the evolution of crack, including cyclic changes, by sensing both the increase and decrease in crack width. The procedure outlined in this paper includes a preliminary investigation into the influence of fibre diameter and density within the patch, which has revealed that fibres of 750 µm in diameter exhibit optimal performance, as well as the attachment method to concrete surfaces. Subsequently, the process of calibration is detailed, followed by numerous laboratory tests that were conducted on reinforced concrete samples of varying sizes. Comparisons and validations of optical fibre patch sensor measurements were performed using Digital Image Correlation, resulting in a sensitivity of the sensors reaching 0.01 mm/K, and an accuracy ranging between 0.1% and 1%. The discussion within this paper considers the installation of this new sensor both externally and internally within real reinforced concrete structures