HAL Portal UPHF (Université Polytechnique Hauts-de-France)
Not a member yet
28473 research outputs found
Sort by
Electricity-driven selectivity in the photocatalytic oxidation of methane to carbon monoxide with liquid gallium-semiconductor composite
No full textInternational audienceMetal co-catalysts in oxide semiconductors are essential to enhance photocatalytic reaction, but the selectivity is often low due to the conversion of the target molecules as more reactive in comparison with the reagents. Here, we explore the influence of electricity on the activity and selectivity of liquid gallium as a co-catalyst loaded in TiO2 and ZSM-5 supports the photocatalytic oxidation of methane. We first show that the polarization of the photocatalytic chips allows a more efficient wetting of liquid gallium into the semiconductor matrices, which increases the availability of the photogenerated charge carriers for reactions and improves the overall activity of the composites. More importantly, the oxide skin surrounding the spread liquid is found to be charged, resulting in a significant reduction of the CO oxidation. An increase of the CO selectivity from 10 to 80 % upon polarization to 50 V, with a stable photocatalytic activity reaching 0.6 ± 0.01 mmol/g•h, makes the use of electricity quite appealing to tailor liquid metals in catalysis
Signal estimation and uncertainties extraction in TeraHertz Time Domain Spectroscopy
No full textInternational audienceTerahertz Time Domain Spectroscopy (THz-TDS) systems have emerged as mature technologies with significant potential across various research fields and industries. However, the lack of standardized methods for signal and noise estimation and reduction hinders its full potential. This paper introduces a methodology to significantly reduce noise in THz-TDS time traces, providing a reliable and less biased estimation of the signal. The method results in an improved signal-to-noise ratio, enabling the utilization of the full dynamic range of such setups. Additionally, we investigate the estimation of the covariance matrix to quantify the uncertainties associated with the signal estimator. This matrix is essential for extracting accurate material parameters by normalizing the error function in the fitting process. Our approach addresses practical scenarios where the number of repeated measurements is limited compared to the sampling time axis length. We envision this work as the initial step toward standardizing THz-TDS data processing. We believe it will foster collaboration between the THz and signal processing communities, leading to the development of more sophisticated methods to tackle new challenges introduced by novel setups based on optoelectronic devices and dual-comb spectroscopy.</div
Paul Demeny et Marceline Desbordes-Valmore, des liens impénétrables ?: À la recherche des "Desbordes" perdus
No full textInternational audience"Dieu bénit l'homme non pour avoir trouvé, mais pour avoir cherché", disait Victor Hugo dans son poème "La vie aux champs". Un chercheur en sciences humaines est souvent amené à faire l'expérience amère de cette sentence hugolienne.Maxance Lardjane, dans le cadre du concours "Show Your PhD", a décidé de faire part d'une quête philologique infructueuse qui a pris neuf mois de son temps : celle d'une pièce de théâtre écrite en 1880 par Paul Demeny, intitulée "Les Desbordes".Il a fait part au public de ses recherches qui, ayant débutées dans les appels de note d'une biographie de Rimbaud, l'ont amené à contacter tous les experts rimbaldiens de France et d'ailleurs. Il aborde étape par étape ce périple durant lequel il a, entre autres choses, obtenu l'acte de décès d'un parent de Paul Demeny, et essayé de contacter un membre lointain de sa famille.Il conclut sur une note ouverte, en disant que même si ses recherches ont été mises temporairement à l'arrêt, il reviendrait bientôt à la charge, non sans avoir défriché le champ des possibles pour plusieurs chercheurs en herbe après lui. Il fait de son expérience un exemple des dérives des chercheurs, parfois prêts à tout pour élucider, en pure perte, le moindre mystère
Problèmes de satisfaction de contraintes : mieux les appréhender par l'interaction tangible, une revue systématique de la littérature
No full textInternational audienceConstraint satisfaction problems (CSPs), whether they involve scheduling, deliveries routing or allocating resources, are ubiquitous in our daily lives. These problems often involve a large number of variables and constraints, making them complex to perceive. At present, they are mainly designed and manipulated via computer code, which adds to this difficulty of appropriation. We believe that a tangible interface (TUI) could facilitate their design and/or manipulation. To explore this hypothesis, we carried out an initial systematic literature review focusing on tangible representations of generic CSPs. However, this revealed the absence of such representations for problems that are not specific to a particular application. As a result, the challenge of realizing a tangible interface representing a generic CSP is unexplored. Consequently, we conducted a second systematic literature review, focusing on tangible interfaces for specialized CSP applications (scheduling, routing, etc.). We extracted from this review forteen articles, which we listed, described, analyzed and compared.Les problèmes de satisfaction de contraintes (CSP), qu'il s'agisse de la planification d'emplois du temps, de la gestion de livraisons ou de l'allocation de ressources, sont omniprésents dans notre quotidien. Ces problèmes impliquent souvent un grand nombre de variables et de contraintes, rendant leurs perceptions complexes. Actuellement, ils sont principalement conçus et manipulés via du code informatique, ce qui ajoute à cette difficulté d'appropriation. Nous pensons qu'une interface tangible (TUI) pourrait faciliter la conception et/ou la manipulation de ces variables et contraintes. Pour explorer cette hypothèse, nous avons réalisé une première revue systématique de la littérature se concentrant sur les représentations tangibles des CSP génériques. Cependant, celle-ci a mis en évidence l'absence de telle représentation pour des problèmes non-spécifiques à une application particulière. Il en résulte donc que le défi de réaliser une interface tangible représentant un CSP générique est inexploré. Par conséquent, nous avons conduit une seconde revue systématique de la littérature, qui s'est attachée à examiner les interfaces tangibles des applications spécialisées de CSP (emploi du temps, gestion de livraison, etc.). Nous avons extrait de cette revue quatorze articles que nous avons listés, décrits, analysés et comparés
RAVID: Retrieval-Augmented Visual Detection: A Knowledge-Driven Approach for AI-Generated Image Identification
No full textIn this paper, we introduce RAVID, the first framework for AI-generated image detection that leverages visual retrieval-augmented generation (RAG). While RAG methods have shown promise in mitigating factual inaccuracies in foundation models, they have primarily focused on text, leaving visual knowledge underexplored. Meanwhile, existing detection methods, which struggle with generalization and robustness, often rely on low-level artifacts and model-specific features, limiting their adaptability. To address this, RAVID dynamically retrieves relevant images to enhance detection. Our approach utilizes a fine-tuned CLIP image encoder, RAVID CLIP, enhanced with category-related prompts to improve representation learning. We further integrate a vision-language model (VLM) to fuse retrieved images with the query, enriching the input and improving accuracy. Given a query image, RAVID generates an embedding using RAVID CLIP, retrieves the most relevant images from a database, and combines these with the query image to form an enriched input for a VLM (e.g., Qwen-VL or Openflamingo). Experiments on the UniversalFakeDetect benchmark, which covers 19 generative models, show that RAVID achieves state-of-the-art performance with an average accuracy of 93.85%. RAVID also outperforms traditional methods in terms of robustness, maintaining high accuracy even under image degradations such as Gaussian blur and JPEG compression. Specifically, RAVID achieves an average accuracy of 80.27% under degradation conditions, compared to 63.44% for the state-of-the-art model C2P-CLIP, demonstrating consistent improvements in both Gaussian blur and JPEG compression scenarios. The code will be publicly available upon acceptance
Non-Destructive Rail Monitoring for Defect Identification
No full textInternational audienceNon-destructive evaluation (NDE) of rail tracks is crucial to ensure the safety and reliability of rail transportation systems. In this work, we present a quantitative study using various signal processing methods to identify defects in rail structures. A diffuse field configuration was employed at few dozens of kiloHertz, where the emitter and receiver were remotely located, and wave energy propagated via multiple reflections within the medium. A reference database is first constructed by acquiring measurements at different rail positions and different torque levels (up to 50 N.m). The defect is then identified by comparing its signature to those stacked in the database. First, the destretching technique, based on Coda Wave Interferometry (CWI), is applied to correct for temperature-induced velocity variations. Then, the identification is performed using the Mean Square Error (MSE) metric and Orthogonal Matching Pursuit (OMP) technique. A comparative analysis of the both methods is conducted, focusing on their robustness and performance.</div
Memory Fusion Sampled-Data Control of Fractional-Order Heterogeneous Multiagent Systems Subject to DoS Attacks and Time Delays: A Resilient Binary Sampled-Data Scheme
No full textInternational audienc
Les actions du CSF électronique pour l’emploi et le développement des compétences
No full textNational audiencePrésentation orale, doublée d'un support poste
The impact of medical status on the choice of dental procedures under general anesthesia—a retrospective study
No full textInternational audienceBackground: The purpose of this study was to assess the type of dental procedures performed on children under General Anesthesia (GA) and to determine if the pattern differs between healthy children and those with special healthcare needs. Methods: In this retrospective study, data were reviewed from the dental records of pediatric patients who underwent dental treatment under GA from 2015 to 2020 at Nantes University Hospital. Patients with mental or physical disabilities were categorized as Disabled (D), while healthy children were assigned to the Healthy group (H). Records from patients with Systemic Diseases were also analyzed with (D + SD) or without (SD) Disabilities. Results: The mean age of each group was evaluated and compared to the others. The number and type of dental treatments were compared between each group for both primary and permanent teeth. A total of 655 patients were treated under GA. Patients in groups H and SD were significantly younger than those in the disabled group (p < 0.001). Primary teeth were more frequently treated in groups H and SD than in groups D and SD + D, while the opposite was true for permanent teeth. There were more extractions of primary teeth than restorative treatments performed in children with disabilities (p = 0.0005). Conclusions: The findings of this study suggest that the health conditions of young patients could impact their dental procedures when undergoing GA. Children with systemic diseases don’t seem to differ from healthy patients in the acts performed, but patients with disabilities do