HAL Portal UO (Université d'Orléans)
Not a member yet
48841 research outputs found
Sort by
Conformal Prediction for LLMs in the Black-Box Model
No full textInternational audienceLes modèles de langage (LM) sont omniprésents dans les tâches de classification de texte, mais pour ceux dite de "boîte noire" l'interprétation de leurs prédictions est grandement limité. Nous introduisons Monte Carlo Jailbreaking (MCJ), une nouvelle méthodologie qui exploite des techniques issues du jailbreaking et d'échantillonnages de type Monte Carlo pour estimer les distributions de probabilité sur les classes de sortie d’un LM.Concrètement, MCJ génère plusieurs réponses du modèle à partir de prompts volontairement modifiés selon différentes stratégies, puis construit empiriquement la distribution des prédictions. Cette estimation permet de quantifier l’incertitude du modèle, de détecter les exemples ambigus ou instables, et d’identifier les cas hors distribution.Nous évaluons plusieurs variantes de MCJ sur des jeux de données de classification standard en analysant la robustesse face à la nuisance des prompts, le calibrage probabiliste des sorties, ainsi que la capacité à repérer les entrées difficiles ou inattendues. Nos expériences démontrent que MCJ constitue un outil robuste pour enrichir l’interprétabilité et renforcer la fiabilité des modèles de langage utilisés en contexte de boîte noire complète
Consistently mapping Differential Privacy paradigms between relational databases and RDF
No full textInternational audienceThe Semantic Web represents an extension of the currentweb offering a metadata-rich environment based on the Resource De-scription Format (RDF) which supports advanced querying and infer-ence. However, relational database (RDB) management systems remainthe most widespread systems for (Web) data storage. Consequently, thekey to populating the Semantic Web is the mapping of RDB to RDF, sup-ported by standardized mechanisms. Confidentiality and privacy repre-sent significant barriers for data owners when considering the translationand subsequent utilization of their data. In order to facilitate acceptance,it is essential to build privacy models that are equivalent, explainable,and usable within both data formats.Differential Privacy (DP) has emerged to be the flagship of data privacywhen sharing or exploiting data. Recent works have proposed DP-modelstailored for either multi-relational databases or RDF. This paper lever-ages this field of work to study how privacy guarantees on RDB withforeign key constraints can be transposed to RDF databases and viceversa.We use classical RDB and RDF formalisms as well as an establishedtranslation tool between the two (RDB2RDF) to compare DP models be-tween the two data representations. We consider a promising DP modelfor RDB related to cascade deletion and demonstrate that it is some-times similar to an existing DP graph privacy model, but inconsistentlyso. Consequently, we tweak this model in the relational world and pro-pose a new model called restrict deletion. We show that it is equivalentto an existing DP graph privacy model, facilitating the comprehension,design and implementation of DP mechanisms in the context of the map-ping of RDB to RDF. Conversely, we consider a useful DP model withlabel differentiation capabilities on graphs (QL-outedge), and propose itstransposition into an original RDB distance
Essais sur la politique fiscale et le développement économique
No full textThis thesis explores the mechanisms by which fiscal policy can foster sustainable and inclusive development in developing countries, with a focus on tax structures, fiscal rules, public spending efficiency, and the fiscal implications of climate shocks and armed conflicts. Through in-depth empirical analysis utilizing advanced econometric methods, this research highlights how well-designed fiscal frameworks, supported by strong institutions, can promote macroeconomic stability, reduce inequality, and improve development outcomes. The first chapter analyzes the impact of tax structures on income inequality in Sub-Saharan African countries. The analysis reveals that progressive personal income tax is the most effective instrument for reducing inequality, especially when the tax system is supported by robust institutions. The findings emphasize the importance of tax reforms that promote equity without compromising internal resource mobilization. The second examines the fiscal challenges faced by Sahel countries affected by armed conflicts. The study reveals that conflicts weaken governments' capacity to mobilize tax revenues, thus reducing the resources available for essential public spending. In chapter three, we explore the impact of climate shocks on public spending efficiency. The findings show that climate change undermines the efficiency of public spending. The chapter emphasizes the urgency of integrating climate resilience into fiscal frameworks to ensure effective public spending management, even in the face of growing environmental crises. Finally, chapter four focuses on the impact of fiscal rules on development outcomes in developing countries. The study demonstrates that countries adopting stringent fiscal rules achieve better outcomes in sectors such as health, education, and infrastructure. These rules strengthen fiscal discipline, allowing for more efficient allocation of public resources, thereby contributing to long-term development. This thesis provides valuable insights into how fiscal policy can be a powerful tool in addressing the key challenges, while fostering sustainable development.Cette thèse explore les mécanismes par lesquels la politique budgétaire peut favoriser un développement durable et inclusif, en mettant l'accent sur les structures fiscales, les règles budgétaires, l'efficience des dépenses publiques et les implications fiscales posés par les chocs climatiques et les conflits armés. À travers une analyse empirique utilisant des méthodes économétriques avancées, nous mettons en évidence comment des cadres fiscaux bien conçus, appuyés par des institutions solides, peuvent favoriser la stabilité macroéconomique, réduire les inégalités et améliorer les résultats en matière de développement. Le premier chapitre analyse l'impact des structures fiscales sur les inégalités de revenus. Il révèle que l'impôt sur le revenu des personnes physiques est l'instrument le plus efficace pour réduire les inégalités, en particulier lorsque le système fiscal est soutenu par des institutions robustes. Les résultats soulignent l'importance de réformes fiscales qui favorisent l'équité sans compromettre la collecte des recettes. Le second chapitre examine les défis fiscaux auxquels sont confrontés les pays du Sahel en proie à des conflits armés internes. L'étude révèle que les conflits affaiblissent la capacité des gouvernements à mobiliser des recettes fiscales, réduisant ainsi les ressources disponibles pour les dépenses publiques essentielles. Dans le troisième chapitre, l'impact des chocs climatiques sur l'efficience des dépenses publiques est exploré. Les résultats montrent que les changements climatiques compromettent l'efficience des dépenses publiques. Le chapitre souligne l'urgence d'intégrer la résilience climatique dans les cadres fiscaux pour maintenir une gestion efficace des dépenses publiques, même face à des crises environnementales croissantes. Enfin, le dernier chapitre s'intéresse à l'impact des règles budgétaires sur les résultats en matière de développement. L'étude démontre que les pays ayant adopté des règles budgétaires obtiennent de meilleurs résultats en matière de santé, d'éducation et d'infrastructures. Ces règles renforcent la discipline budgétaire, ce qui permet une allocation plus efficace des ressources publiques et contribue au développement à long terme. Cette thèse fournit des indications précieuses sur la manière dont la politique fiscale peut être un outil puissant pour relever les défis clés tout en favorisant le développement durable
Simulation Models for Exploring Magnetic Reconnection
No full textInternational audienceSimulations have played a critical role in the advancement of our knowledge of magnetic reconnection. However, due to the inherently multiscale nature of reconnection, it is impossible to simulate all physics at all scales. For this reason, a wide range of simulation methods have been crafted to study particular aspects and consequences of magnetic reconnection. This article reviews many of these methods, laying out critical assumptions, numerical techniques, and giving examples of scientific results. Plasma models described include magnetohydrodynamics (MHD), Hall MHD, Hybrid, kinetic particle-in-cell (PIC), kinetic Vlasov, Fluid models with embedded PIC, Fluid models with direct feedback from energetic populations, and the Rice Convection Model (RCM)
An Evaluation of the Effect of Dimple Insoles on Foot Temperature in Diabetic Patients
No full textInternational audienceObjective: Insoles play a crucial role in foot comfort, with their effect on foot temperature being a key factor. This study aims to evaluate and compare the effect of walking with two different insole types—dimple insoles versus a conventional insole—on foot temperature changes in patients with diabetic neuropathy. Methods: Thermal imaging was used to measure the foot temperature of nine participants immediately before and after walking 250 m in each insole. Temperature variations were analyzed for the whole foot across four specific regions to assess and compare the effect of each insole on foot temperature. Results: The Wilcoxon Signed-Rank Test revealed that contralateral temperature differences between the left and right feet after walking (∆TAfter) were significantly (p<0.05) lower in dimple insoles compared to the conventional insoles. This effect was particularly strong in the midfoot and toe regions. Conclusions: The results indicate that insole type can influence foot contralateral temperature differences after walking. These findings provide valuable insights for selecting insoles based on thermal data and can have implications in improving patient outcomes
Technology Demonstration testing of the PPS-20k Thruster
No full textInternational audienceWithin the framework of CHEOPS very high-power building blocks programme. Safran tested its 20 kW Hall thruster, the PPS-20k, after more than 13 years since the last firing in 2012. The thruster, equipped with a new high current cathode and ceramic, was tested in Xe, Kr and Ar, showing excellent performance on Kr and promising results on Ar. The test features several plasma diagnostics to increase the thruster behavior with the three propellants in the view of the next developments and qualifications activities
Optical Emission Spectroscopy and Laser-Induced Fluorescence on iodine plasma discharges for electric propulsion purposes
No full textInternational audienc
The Role of Perceived Residual Value in Post-Consumer Behaviours
No full textInternational audienceThis exploraty study aims to deepen the understanding of the role played by the perceived residual value of everyday consumer goods, with a focus on cell phones and jewelry in post-consumer decision-making. Through a comparative qualitative analysis (N=19), this research examines how the perceived residual value of these items influences consumers' decisions regarding their disposal or reuse. By replicating and extending the foundational work of Kreziak et al. (2016) to a new context, the study sheds light on the key dimensions of perceived residual value and its effects on the end-of-life fate of products.Keywords: Perceived Value, Perceived Residual Value, Sustainable Consumption, Post-Consumer Behavior, Qualitative Study, Post-Use Behavio
Une paramétrisation stochastique du mélange induit par les ondes de gravité non orographiques pour le Modèle Climatique Planétaire de Mars
No full textInternational audienceThis paper presents a formalism of mixing induced by non‐orographic gravity waves (GWs) to integrate with the stochastic GWs scheme in the Mars Planetary Climate Model. We derive the formalism of GWs and their mixing under the same assumptions, integrating the two schemes within a unified framework. Specifically, a surface‐to‐exosphere parameterization of GW‐induced turbulence has been derived in terms of the eddy diffusion coefficient. Simulations show that the coefficient is on the order of 10⁴ to 10⁹ cm² s⁻¹ and a turbopause is at altitudes of 70–140 km, varying with seasons. The triggered mixing has minor effects on model temperatures, yet it substantially impacts upper atmospheric abundances. Simulations are consistent with observations from the Mars Climate Sounder and the Neutral Gas and Ion Mass Spectrometer. Mixing enhances the tracer transports in the middle and upper atmosphere, governing the dynamics of these regions. The scheme reveals how non‐orographic GW‐induced turbulence can regulate upper atmospheric processes, such as tracer escape.Cet article présente un formalisme du mélange induit par les ondes de gravité (OG) non orographiques afin de l’intégrer au schéma stochastique des OG dans le Modèle Climatique Planétaire de Mars. Nous dérivons le formalisme des OG et de leur mélange sous les mêmes hypothèses, en intégrant les deux schémas dans un cadre unifié. Plus précisément, une paramétrisation de la turbulence induite par les OG, de la surface à l’exosphère, a été formulée en termes de coefficient de diffusion turbulente. Les simulations montrent que ce coefficient est de l’ordre de 10⁴ à 10⁹ cm² s⁻¹ et qu’une turbopause se situe entre 70 et 140 km d’altitude, variant selon les saisons. Le mélange ainsi déclenché a un effet limité sur les températures du modèle, mais il influence fortement les abondances de la haute atmosphère. Les simulations sont cohérentes avec les observations du Mars Climate Sounder et du Neutral Gas and Ion Mass Spectrometer. Le mélange renforce le transport des traceurs dans la moyenne et la haute atmosphère, gouvernant la dynamique de ces régions. Ce schéma met en évidence la façon dont la turbulence induite par les OG non orographiques peut réguler les processus de la haute atmosphère, tels que l’échappement des traceurs