HAL: Hyper Article en Ligne
Not a member yet
3159010 research outputs found
Sort by
Un maccarthysme français ? L'exclusion de scientifiques pendant la Guerre froide
No full textInternational audienc
Thermal and mechanical behaviors of optical silica glass fiber during the drawing process
No full textInternational audienceThe process of fiber drawing in a vertical furnace is described using the lubrication approximation coupled to heat transfer. The radiative heat transfer is detailed by studying the emitted and absorbed fluxes experimented by the fiber. The emissivity is carefully determined with high spectral resolution. Using a two-band gray absorption coefficients, the Planck average emissivity is overestimated by 24 % in comparison with a high spectral resolution of absorption coefficient. The profile of the heating area is obtained from the experimental data. The steady state lubrication model is solved numerically using a finite difference method. A numerical prediction obtained under certain operating conditions allows a comparison of the fiber radius profiles obtained experimentally and numerically. The convective heat transfer after the neck-down region is needed to control the cooling and the fiber shape. Drawing forces obtained experimentally are compared to numerical results. When the operating temperature is equal to 1950 • C, the agreement is satisfying. The effects of variations in drawing velocity and furnace wall temperature on the relevant operating conditions for adjusting the cooling rate, drawing force, and other variables are numerically studied.</div
Leveraging model explainability and fine-grained cutmix augmentation for robust detection of apricot diseases in UAV images
No full textInternational audienceApricots (Prunus Armeniaca) are valuable stone fruits cultivated worldwide in temperate regions, generating $500 million in annual exports. However, disease and pests significantly threaten apricot production, impacting quality and yield. Brown rot and shot hole are the two major diseases affecting apricot yield worldwide. Early detection and targeted management strategies are critical to prevent their spread. Unfortunately, the lack of diverse labeled datasets hinders the performance of deep learning models for in-field disease detection. In this regard, we propose an innovative approach that leverages deep convolutional generative adversarial networks (DCGAN) and model-guided Cutmix (MGC) data augmentation to synthesize images of diseased apricots. The proposed synthesis process is driven by counterexamples, which are samples that the model fails to detect correctly. The use of DCGAN for background and conditional DCGAN for foreground generation allows fine-grained control over the synthesis of new samples. Further, the MGC uses SHapley Additive exPlanations of the object detection model to analyze its weaknesses and guide on-demand sample generation by replicating the counterexample’s label distribution, aspect ratio, scale, and brightness/contrast. This expands and diversifies our custom apricot disease dataset, initially containing 1500 images. This addresses dataset imbalances and exposes the model to a dynamically augmented dataset, iteratively improving its performance. The proposed method was extensively evaluated on images of healthy and diseased apricots captured by unmanned aerial vehicles in different environmental conditions. MGC outperformed traditional augmentation methods with even smaller dataset, achieving 4 % better mean average precision (mAP) score with Apricot-3K dataset. Additionally, the proposed method achieved 1–4 % improvements in mAP with many state-of-the-art object detection models when trained using the proposed framework on the Apricot-10K augmented dataset
Development of a numerical model for measuring the electrical conductivity (EC) of a cake batter
No full textInternational audienceThe electrical conductivity (EC) of materials represents their ability to conduct electrical current and determines the power dissipated within the material. This parameter can be temperature and electric field dependent. When EC is measured in the electric fields used for ohmic heating, the increase in temperature results in heating nonuniformity in the measuring cell. As a result, the relationship between EC and temperature cannot be accurately determined. In addition, for cake batter, starch gelatinization occurs during the measurement, leading to more complex EC curves. To address these issues, this study proposes a numerical method for determining EC that accounts for temperature non-uniformity. This model is coupled with a starch gelatinization model. The principle is based on the estimation of the EC using the method of least squares between the experimental temperature and the numerical one. The estimation of the EC of the cake batter consists of two steps: first, the device was characterized with xanthan and potassium chloride solutions of known electrical conductivities. A conversion efficiency of 0.77 was found. This efficiency was used to estimate the EC of the cake batter as a function of temperature and for different electric fields. Results showed that EC became independent of the electric field from 34.57 V/cm
Le triomphe de Robert Dudley à La Haye (1586): la représentation graphique d'une "scène de guerre" à l'issue incertaine?
No full textInternational audienc
The High Voltage Splitter board for the JUNO SPMT system
No full textInternational audienceThe Jiangmen Underground Neutrino Observatory (JUNO) in southern China is designed to study neutrinos from nuclear reactors and natural sources to address fundamental questions in neutrino physics. Achieving its goals requires continuous operation over a 20-year period. The small photomultiplier tube (small PMT or SPMT) system is a subsystem within the experiment composed of 25600 3-inch PMTs and their associated readout electronics. The High Voltage Splitter (HVS) is the first board on the readout chain of the SPMT system and services the PMTs by providing high voltage for biasing and by decoupling the generated physics signal from the high-voltage bias for readout, which is then fed to the front-end board. The necessity to handle high voltage, manage a large channel count, and operate stably for 20 years imposes significant constraints on the physical design of the HVS. This paper serves as a comprehensive documentation of the HVS board: its role in the SPMT readout system, the challenges in its design, performance and reliability metrics, and the methods employed for production and quality control
Operational experience and performance of the Silicon Vertex Detector after the first long shutdown of Belle II
No full textInternational audienceIn 2024, the Belle II experiment resumed data taking after the Long Shutdown 1, which was required to install a two-layer pixel detector and upgrade accelerator components. We describe the challenges of this shutdown and the operational experience thereafter. With new data, the silicon-strip vertex detector (SVD) confirmed the high hit efficiency, the large signal-to-noise ratio, and the excellent cluster position resolution. In the coming years, the SuperKEKB peak luminosity is expected to increase to its target value, resulting in a larger SVD occupancy caused by beam background. Considerable efforts have been made to improve SVD reconstruction software by exploiting the excellent SVD hit-time resolution to determine the collision time and reject off-time particle hits. A novel procedure to group SVD hits event-by-event, based on their time, has been developed using the grouping information during reconstruction, significantly reducing the fake rate while preserving the tracking efficiency. The front-end chip (APV25) is operated in the multi-peak mode, which reads six samples. A 3/6-mixed acquisition mode, based on the timing precision of the trigger, reduces background occupancy, trigger dead-time, and data size. Studies of the radiation damage show that the SVD performance will not seriously degrade during the lifetime of the detector, despite moderate radiation-induced increases in sensor current and strip noise
Aomar Hannouz, Le cycle de ʿAbd al-Muṭṭalib. Restauration et naissance prophétique dans laSīra d’Ibn Isḥāq, Brill, Leyde, Boston, 2024
No full textInternational audienc
Émergence d’un rôle frontière au sein d’un programme national piloté par Santé Publique France : le cas des chargé.es d’étude du programme national de surveillance du mésothéliome pleural
No full textInternational audienc
Stochastic quay partitioning problem
No full textInternational audienceIn this paper we consider the problem of dividing a quay of a container terminal into berth segments so that the quality of service for future ship arrivals is as good as possible. Since future arrivals are unknown, the alternative solutions are evaluated on various arrival scenarios generated for certain arrival intensity from a stochastic model referred to as a ship traffic model (STM). This problem will be referred to as a stochastic quay partitioning problem (SQPP). SQPP is defined by an STM, arrival intensity, quay length and a set of admissible berth lengths. Evaluation of an SQPP solution on one scenario is a problem of scheduling the arriving vessels on the berths, which is a classic berth allocation problem (BAP). In SQPP the sizes of BAP instances that must be solved by far exceed capabilities of the methods presented in the existing literature. Therefore, a novel approach to solving BAP is applied. Tailored portfolios of algorithms capable of solving very large BAP instances under limited runtime are used. Features of SQPP solutions are studied experimentally: patterns in selected berth lengths, dispersion of solutions quality and solutions similarity. We demonstrate, that partitioning a quay into equal-length berths is not the best approach. The largest vessel traffic is dominating in defining best quay partitions, but dedicating quays for shorter vessels give lower dispersion of solution quality. A set of algorithms to partition a quay is proposed and evaluated: methods based on integer linear programming (ILP) to match vessel classes arrival intensities with berth availability, hill climber, tabu search and a greedy approach. Only under high arrival intensity can these methods show their prowess. ILP methods have an advantage of low solution evaluation cost. Tabu is most flexible, but at high evaluation costs. To the best of our knowledge, SQPP is posed and solved for the first time in the operations research