HAL Evry
Not a member yet
19237 research outputs found
Sort by
Machine Learning-driven Optimization of a Sensor Network for Accurate Pollutant Source Identification
No full textInternational audienceOptimizing sensor networks for localizing atmospheric pollution sources and enhancing estimation accuracy remains a significant challenge in air pollution studies. To address this, various techniques have been recently developed. Among them, machine learning has demonstrated its ability to model and optimize complex problems, including sensor network optimization.To improve the localization of atmospheric pollution sources in air quality research activities, we propose in this paper, a Machine Learning-driven Optimization of Sensor Networks method (ML-OSN). The method introduces a new combination of Hierarchical Agglomerative Clustering and Siamese Neural Networks, thereby improving the prediction of similarities in pollutant concentrations across different wind directions and leading to an optimized sensor network.The proposed ML-OSN method was evaluated and compared to a standard clustering approach based on the Pearson correlation coefficient, using the augmented Indianapolis dataset. The resulting optimal sensor network configuration achieved broader spatial coverage and improved source estimation accuracy, reducing the error score to 1.34 compared to 1.44 obtained with the Pearson-based approach
An overview on simple P systems with specific variants of derivation modes
No full textInternational audienceIn this paper, we consider simple P systems, i.e., membrane systems with only one membrane region, together with various derivation modes as they have been defined and thoroughly investigated during the last 25 years, for instance, the variants of parallel derivation modes using non-extendable multisets of rules as the maximally parallel derivation mode, but also some more complicated variants, for example, simple P systems with prescribed teams of sets of rules as introduced recently. In this context, we recall many known results, especially for purely catalytic and catalytic P systems. In most cases, computational completeness of these simple P systems has been established, in other cases, especially when using only non-cooperative rules, only regular sets can be obtained, and in some cases, we get the same computational power as ET0L systems used for multisets
Diagnosis accuracy of touchscreen-based testings for major neurocognitive disorders: a systematic review and meta-analysis
No full textInternational audienceFacing an increasing prevalence, diagnosis and management of dementia has become a global health challenge. While most cognitive assessments are based on paper tests, the current trend is to digitise them, using new technologies. We aimed to achieve a systematic review on touchscreen-based assessments for neurocognitive disorders. A search on four databases (PubMed, Embase, IEEE Xplore and Web of Science) was performed. Two investigators independently selected the articles and assessed their quality using the QUADAS-2 tool. We included articles whose participants were over 65 years, classified according to the presence/absence of major neurocognitive disorder (M-NCD) determined by conventional assessment of cognition, examined with a novel tool using a touchscreen device (tablet or smartphone). We finally retained 35 articles for the systematic review and 24 for the meta-analysis. Of the 35 articles included in the systematic review, participants had Alzheimer’s disease, Parkinson’s disease, vascular dementia, Lewy body disease or any type of dementia. Pooled sensitivity and specificity were 0.89 (95% CI: 0.86–0.91) and 0.88 (95% CI: 0.82–0.92), respectively. Performances of self-administered or brief assessment testings were similar to those of assessor-administered or longer duration testings. The major limitation of this review and meta-analysis is the multiplicity and diversity of methods (tools, cognition assessment and type of M-NCD), which make the comparison difficult. We conclude that brief and self-assessment touchscreen-based cognitive tests are appropriate and simple tools to screen major cognitive disorders that can be used in primary care. The study was registered in Prospero (CRD42022358725)
Control for Singular Fractional-Order Systems
No full textInternational audienceAs pointed out in Chap. 4, considerable efforts have been devoted to stabilization for fractional-order systems. Over the last years singular values have been developed as a tool to deal with this problem. In general terms, the controller should be chosen so that the closed-loop transfer function matrix has certain characteristics that are derived from the specifications. An optimal design minimizes the maximum singular value of the desired loop shape, subject to a closed-loop stability constraint
Text-driven Motion Generation: Overview, Challenges and Directions
No full textText-driven motion generation offers a powerful and intuitive way to create human movements directly from natural language. By removing the need for predefined motion inputs, it provides a flexible and accessible approach to controlling animated characters. This makes it especially useful in areas like virtual reality, gaming, human-computer interaction, and robotics. In this review, we first revisit the traditional perspective on motion synthesis, where models focused on predicting future poses from observed initial sequences, often conditioned on action labels. We then provide a comprehensive and structured survey of modern text-to-motion generation approaches, categorizing them from two complementary perspectives: (i) architectural, dividing methods into VAE-based, diffusion-based, and hybrid models; and (ii) motion representation, distinguishing between discrete and continuous motion generation strategies. In addition, we explore the most widely used datasets, evaluation methods, and recent benchmarks that have shaped progress in this area. With this survey, we aim to capture where the field currently stands, bring attention to its key challenges and limitations, and highlight promising directions for future exploration. We hope this work offers a valuable starting point for researchers and practitioners working to push the boundaries of language-driven human motion synthesis
Admissibility Analysis of Singular Fractional-Order Systems
No full textInternational audienceResearch on dynamic systems often necessitates the mathematical modeling of system behavior to understand and predict their performance. A significant class of physical systems can be represented using Differential-Algebraic Equations (DAEs), commonly referred to as singular systems. This chapter focuses on a novel category of singular linear systems characterized by the presence of non-integer order derivatives, offering a fresh perspective on the analysis and control of such systems
Les effecteurs du nématode à galles ciblent le mécanisme d'épissage alternatif des plantes afin de permettre la formation de cellules géantes.
No full textInternational audienceRoot-knot nematodes (RKNs) are obligate endoparasites that establish a biotrophic relationship with host plants. To manipulate root cells, they secrete effectors from their esophageal glands that are delivered into the host via a stylet. These effectors target distinct subcellular compartments and induce the redifferentiation of root cells into multinucleate giant cells essential for nematode development. We characterised MiEFF18, a nuclear-targeted effector that accumulates in the host cell nucleolus. A yeast two-hybrid screen identified its interaction with a core nuclear spliceosomal protein, SmD1. Functional analyses revealed the critical role of this target in plant susceptibility to RKNs, specifically in giant cell formation. We showed that MiEFF18 alters alternative splicing, a key post-transcriptional process that generates multiple mRNA isoforms from a single gene. A complementary screening approach identified MiEFF186 as an additional effector involved in the regulation of alternative splicing. MiEFF186 interacts with a tomato splicing factor, further highlighting the role of RKN effectors in manipulating host gene expression. This study reveals a novel nematode strategy for subverting plant nuclear and nucleolar functions and alternative splicing mechanisms, and provides insights into how RKNs exploit host cellular machinery to establish and maintain their feeding sites
Utilizing plantar pressure perception for teleoperation: enhancing the control of humanoid robot wheeled chassis
No full textInternational audienceTo enhance the response sensitivity and user interaction experience in the teleoperation control of a humanoid robot’s wheeled chassis, this paper proposes a teleoperation control system based on plantar pressure perception shoes and introduces a dynamic region boundary adjustment strategy. This strategy increases the absolute value of the pressure difference between the front and rear feet regions, allowing users to achieve significant control effects with small movements. We designed and conducted experiments for polynomial fitting and analysis, and control strategy validation, ultimately determining the specific model of the mapping function. Experimental results show that after dynamic boundary adjustment, the system achieves a greater absolute value of the pressure difference, providing higher linear velocity output under the same mapping conditions. This makes the system more responsive and significantly improves the user interaction experience
Disease exacerbation in human DMD MYOrganoids enables gene therapy evaluation and unveils persistence of fibrotic activity
No full textABSTRACT Duchenne muscular dystrophy (DMD) is a lethal muscle wasting disease caused by absence of dystrophin, a protein essential to preserve muscle integrity continuously challenged by contractions. Gene therapy utilizing adeno-associated virus (AAV) to deliver truncated forms of dystrophin (µDys) is currently the most promising therapeutic approach. However, the therapeutic outcome in treated patients has not been as successful as anticipated by animal studies, underscoring the need of improved and high-throughput models for accurate and fast prediction of human response. Here, we describe the generation of MYOtissues, a 3D muscle platform based on direct myogenic conversion of human induced pluripotent stem cells (iPSC), whose structural and functional maturation is enhanced by fibroblasts incorporation. MYOtissues derived from DMD-iPSC including DMD fibroblasts, exacerbated pathogenic hallmarks such as fibrosis and muscle force loss. As a proof of concept, we showed that AAV-mediated µDys gene transfer improved muscle resistance and membrane stability in DMD-MYOtissues, highlighting the suitability of our system for gene therapy screening
