Archives ouvertes de Paris-Saclay
Not a member yet
272730 research outputs found
Sort by
Review article: Weddell Sea Polynya formation, cessation and climatic impacts
No full textInternational audienceOpen-ocean polynyas, areas with little or no sea ice, reappeared extensively in 2016 and 2017 over the Maud Rise in the Weddell Sea after a 40-year hiatus, raising a series of unresolved questions about the atmosphere-ice-ocean interactions in the Antarctic region. These major polynyas significantly influence moisture and heat exchange between the atmosphere and the ocean, impacting both regional and global climate dynamics, as well as ecosystem functioning and biogeochemical processes. Notably, they may play a crucial role in contributing to the formation of Antarctic Bottom Water and influencing global ocean circulation. In this Review, we synthesize current knowledge on the drivers and impacts of Weddell Sea polynyas. Recent occurrences have been linked to factors such as a strengthening Weddell Gyre, a negative Southern Annular Mode, extreme local atmospheric conditions (atmospheric rivers and cyclones), and subsurface ocean heat buildup which acts as a preconditioning factor. The associated deep ocean convection from these polynyas can enhance air-sea gas exchange and trigger earlier phytoplankton blooms due to the influx of iron and nutrients from the deep ocean. While advancements in observation and modeling techniques have significantly improved our understanding of polynyas, substantial uncertain-ties remain regarding their interaction with recent Antarctic sea ice loss, their sensitivity to ocean mixing schemes, their excessive size or frequency in climate simulations, and future projections. Therefore, future research should focus on developing comprehensive four-dimensional regional observatories and targeted, data-constrained coupled models that accurately capture atmosphere-ice-ocean interactions across various timescales
Electrochemical investigation of nickelolactones in olefin carboxylation reactions
No full textInternational audienc
Token Sample Complexity of Attention
No full textAs context windows in large language models continue to expand, it is essential to characterize how attention behaves at extreme sequence lengths. We introduce token-sample complexity: the rate at which attention computed on tokens converges to its infinite-token limit. We estimate finite- convergence bounds at two levels: pointwise uniform convergence of the attention map, and convergence of moments for the transformed token distribution. For compactly supported (and more generally sub-Gaussian) distributions, our first result shows that the attention map converges uniformly on a ball of radius at rate , where grows exponentially with . For large , this estimate loses practical value, and our second result addresses this issue by establishing convergence rates for the moments of the transformed distribution (the token output of the attention layer). In this case, the rate is with β<\tfrac{1}{2}, and depends polynomially on the size of the support of the distribution. The exponent depends on the attention geometry and the spectral properties of the tokens distribution. We also examine the regime in which the attention parameter tends to infinity and the softmax approaches a hardmax, and in this setting, we establish a logarithmic rate of convergence. Experiments on synthetic Gaussian data and real BERT models on Wikipedia text confirm our predictions
Détermination expérimentale de l’énergie de fissuration de bétons coquillers par corrélation d’images numériques
No full textInternational audienceDans le cadre de la réduction de l'empreinte environnementale des bétons, de récentes études ont permis de montrer qu'il était possible de substituer 50% des gravillons naturels par des co-produits coquilliers d'huîtres, tout en conservant des propriétés mécaniques suffisantes pour des bétons porteurs et en améliorant leurs propriétés de durabilité. Cependant, l'impact de ces granulats alternatifs sur la fissuration des bétons n'a pas encore été étudié. Le travail présenté ici s'intéresse à la fissuration et plus particulièrement à l'énergie de fissuration de bétons incluant des co-produits coquilliers. Des essais de flexion trois points sur éprouvettes entaillées ont été réalisés. L'utilisation de la corrélation d'images numériques (CIN) a permis de mesurer les champs de déplacement et d'observer la propagation de la fissure au cours de l'essai. En particulier, une méthode utilisant l'identification des séries de Williams par méthode intégrée a été utilisée. Elle permet de déterminer les facteurs d'intensité du matériau et de déduire l'énergie de fissuration pour chaque éprouvette
Shift your Focus for the Greater Good: Improving Fairness at no cost for Accuracy and Diversity in News Recommender Systems
No full textInternational audienceIn today’s digital landscape, recommender systems assist users in navigating the vast amount of available data. Within the realm of information access, a subset of such systems called News Recommender Systems help users find news content that interests them. However, by prioritizing traditional accuracy-focused optimization, these systems contribute to the formation of filter bubbles, restricting users’ exposure to diverse viewpoints and exacerbating polarization. To address this issue, beyond-accuracy factors like diversity have been integrated into recommendation. Yet, such approaches can be ineffective and even unintentionally influence user opinions. This raises major ethical concerns as systems lack the legitimacy to shape opinions. This paper presents the ADF framework, a novel approach designed to optimize accuracy, diversity, and fairness simultaneously. Unlike conventional models that manage fairness in a trade-off, ADF establishes fairness as a core constraint. The framework relies on an innovative fairness-constrained diversification strategy, ensuring that users are exposed to a broader range of opinions, without being oriented towards specific viewpoints. ADF is adaptable to various diversity metrics and provides personalized diversification, independent of the underlying recommendation algorithms. Through real-world benchmark datasets and multiple recommendation models, experimental evaluation confirmed that ADF limits the impact on accuracy, enhances diversity, and crucially upholds fairness
RAM-VQA: Restoration Assisted Multi-Modality Video Quality Assessment
No full textInternational audienceVideo Quality Assessment (VQA) strives to computationally emulate human perceptual judgments and has garnered significant attention given its widespread applicability. However, existing methodologies face two primary impediments:(1) limited proficiency in evaluating samples at quality extremes (e.g., severely degraded or near-perfect videos), and (2) insufficient sensitivity to nuanced quality variations arising from a misalignment with human perceptual mechanisms. Although vision-language models offer promising semantic understanding, their reliance on visual encoders pre-trained for high-level tasks often compromises their sensitivity to low-level distortions. To surmount these challenges, we propose the Restoration-Assisted Multi-modality VQA (RAM-VQA) framework. Uniquely, our approach leverages video restoration as a proxy to explicitly model distortion-sensitive features. The framework operates through two synergistic stages: a prompt learning stage that constructs a quality-aware textual space using triple-level references (degraded, restored, and pristine) derived from the restoration process, and a dual-branch evaluation stage that integrates semantic cues with technical quality indicators via spatio-temporal differential analysis. Extensive experiments demonstrate that RAM-VQA achieves state-of-the-art performance across diverse benchmarks, exhibiting superior capability in handling extremequality content while ensuring robust generalization.</div
Stability criteria for singularly perturbed impulsive linear switched systems
No full textWe study a class of singularly perturbed impulsive linear switched systems exhibiting switching between slow and fast dynamics. To analyze their behavior, we construct auxiliary switched systems evolving in a single time scale. We prove that the stability or instability of these auxiliary systems directly determines that of the original system in the regime of small singular perturbation parameters
A practitioner’s framework for reorienting plant health surveillance toward agroecological outcomes
No full textEpidemiological surveillance (ES) for crop production is the systematic process of collecting, analyzing and interpreting plant health data with the purpose of providing information for decision-making concerning practices that can avoid, mitigate or cure disease. At present, converging social and environmental concerns are increasing pressure on farmers and land managers to reduce or even phase out synthetic pesticides, creating a pressing need to better align ES with low-input, ecological-based crop production practices. Here we propose a framework for conceiving and prioritizing indicators for ES based on their potential to contribute specifically to management strategies that are curative or eradicative, prophylactic or that are based on agroecological principles. The framework organizes indicators along eight spatiotemporal axes and across different types of observations (explicit and collateral manifestations of pathogens and diseases, and predisposing factors), and it characterizes their contribution to each of the three management practices. Based on synthesis of the literature and expert knowledge for several diseases and syndromes in perennial and annual crops as illustrative case studies, we show how current epidemiological knowledge can inspire upgraded surveillance strategies that better support agroecology, and we point to future trends where additional research and tools are needed. This framework is novel because it goes beyond recent calls for modernizing ES that generally emphasize technological innovations, yet remain largely focused on observing pests and diseases without clear explanation of the associated management decisions nor the type of agricultural production system (conventional, organic, etc.) they target
Exploring the application of Earth Observation datasets for SEEA carbon accounting and its comparison with national GHG reporting to the UNFCCC
No full textInternational audienceGlobal biomass and carbon datasets derived from Earth Observation (</div
Proposition of an ontology supporting context modeling in operational scenario development
No full textInternational audienceAir Traffic Control systems are complex sociotechnical systems which do not comprise solely technical elements, but also have human and organizational dimensions. Recent approaches to Systems Engineering, and in particular Human Systems Integration, are placing increasing emphasis on incorporating these non-technical considerations into the design processes of complex sociotechnical systems. In particular, one crucial aspect in design is to understand the operational context of the system. There is a lack of tools and languages expressive enough to include contextual information into system models. This is especially true for the Air Traffic Control field, as the operational context of a control tower and the tower itself are deeply intertwined and interdependent. This paper is a step towards the achievement of better integration of context-related knowledge into system design processes. We conducted a case study analysis based on a literature review and feedback from civilian and military air traffic control practitioners, and we propose an ontology of the contextual elements that characterize the context of air traffic control operations