Archive ouverte de Centrale Lyon
Not a member yet
32420 research outputs found
Sort by
Past2ECO : Des cultures du passé aux cultures du futur : utilisation de solutions numériques pour mobiliser la diversité génétique adaptative pour l'agroécologie.
No full textNational audienc
CCASL: Counterexamples to comparative analysis of scientific literature—Application to polymers
No full textInternational audienceThe exponential growth of scientific publications has made the exploration and comparative analysis of scientific literature increasingly complex. For instance, identifying pairs of publications that diverge on widely accepted concepts within a domain is extremely difficult, if not impossible, at a large scale. Our work aims to automatically detect such discrepancies using recent artificial intelligence techniques. Given a particular scientific domain, we propose to capture domain knowledge through the definition of arbitrary functions expressed as relaxed functional dependencies (RFDs), and then focus on the large-scale analysis of tables in the publications related to these RFDs. In this context, we propose a four-step method called Counterexamples to Comparative Analysis of Scientific Literature (CCASL), which consists of the following steps: (1) Modeling the domain knowledge with functions expressed as RFDs, (2) Acquiring a corpus of related publications, (3) Analyzing all tables in the PDF documents and producing a consolidated table, (4) Detecting counterexamples of the RFDs in the consolidated table and conducting a comparative analysis of the pairs of papers containing the detected counterexamples. We have applied CCASL to a subfield of polymer research by identifying an arbitrary function relating the storage modulus, the polymer structure, and the glass transition temperature. Based on this function, we implemented the four steps of CCASL for largescale bibliographic confrontation in polymer science, which enabled us to detect several counterexamples. After detailed analysis, these counterexamples were found to originate from two main sources: typographical errors and methodological inconsistencies. The latter led to an update of the initial arbitrary function, specifying that it is valid only for fully reacted mixtures
Hybrid nonlinear energy sink, a tunable device for vibration mitigation : Theorical and experimental demonstration
No full textInternational audienceHybridization is a powerful and recent technology that combines passive and active componentsto enhance system performance. The central idea behind this approach is to improve the respon-siveness and adaptability of the passive energy absorption mechanism by integrating a control lawsupported by actuator/sensor pairs. This combination enables the system to dynamically adapt it-self to varying excitation conditions, thereby significantly improving overall performance. Mean-while, nonlinear energy sinks (NES) are well known for their frequency robustness. The weakpoint of a nonlinear energy sink lies in its activation threshold, i.e.: the absorber is inactive atlow or high levels of external force applied to the primary structure. Additionally, under a sig-nificant shift in the primary structure’s natural frequencies, the absorber becomes ineffective orinactive. Furthermore, the reactivity of passive absorbers may be insufficient depending on theapplication. For these reasons, a new device for vibration mitigation called the hybrid nonlinearenergy sink is introduced. The latter is designed to overcome the limitations of passive linearand nonlinear absorbers and hybrid linear absorbers. It enhances the non-linear passive energysink by integrating sensor-actuator pairs and a specially designed control law to modify the sys-tem dynamics in real time. This ensures that the absorber remains effective even when externalconditions change. This paper presents theoretical and experimental demonstrations of this newdevice, applied to a one-storey building. By introducing control laws within the hybrid nonlinearenergy sink, it becomes possible to adjust its parameters, e.g., the nonlinear stiffness, to shift theactivation threshold and ensure consistent performance of the absorber in real-time
Air Gaps Fabrication for Sub-100 nm GaN HEMTs by Novel SF6 Plasma Etching
No full textInternational audienceWe demonstrate the fabrication of air gaps in a PECVD SiN interlayer through lateral recess by employing two consecutive plasma etch steps on an AlN/SiN/Al2O3 stack. This approach enables the preservation of sub-100 nm openings in Al2O3, offering a potential optimization for the GaN-HEMT gate stack in RF applications while retaining low gate foot dimensions. A low-power, SF6-based plasma etch is introduced, and time-dependent etch profiles reveal the formation of a skirt-like profile. The process exhibits excellent selectivity between SiN and Al2O3 etch rates. Furthermore, low-power SF6 plasma produces a small self-bias voltage, and surface fluorine contamination which can subsequently be eliminated by annealing
Soil image classification and segmentation: A survey from deep learning, multimodal data and hybrid models
No full textInternational audienceThis survey explores recent advances in the automated analysis of soil images, with a focus on classification and segmentation techniques driven by machine learning and deep learning. The paper provides a comprehensive overview of methods ranging from traditional approaches, such as handcrafted descriptors and statistical classifiers to state-of-the-art neural architectures, including Convolutional Neural Networks, transformers and hybrid multimodal models. While early methods relied heavily on texture and color features, modern deep learning approaches demonstrate enhanced performance through end-to-end learning and the ability to capture complex spatial patterns in soil images. The survey distinguishes between different classification paradigms : object detection, pixel-wise segmentation and general category classification. It emphasizes their relevance for soil analysis tasks such as identifying soil types, mineral compositions or predicting soil properties from RGB or spectral data. Self-supervised and transfer learning techniques are highlighted as promising solutions to the challenge of limited annotated datasets. Beyond classification, the paper discusses the critical role of segmentation in analyzing soil grain morphology, granulometry and spatial distribution. Deep learning-based segmentation methods offer significant improvements over traditional image processing techniques, particularly in heterogeneous or occluded conditions. These techniques enable more accurate particle size estimation and detailed soil texture analysis, information essential for geotechnical, hydrological and environmental applications. Increasingly, hybrid, multimodal and segmentation-aware approaches are being explored, not as an endpoint, but as the next frontier in addressing current analytical limitations to enhance robustness and generalization
Séries génératrices de matroïde
No full textGiven a finite set - for instance, the set of edges in a finite multigraph - it is customary to study the poset of its subsets and, in particular, its incidence algebra, which forms a subalgebra of the algebra of upper triangular matrices. As a vector space, this algebra is spanned by pairs of subsets satisfying . Notably, there is a bijection between these pairs and the minors of a matroid (or graph) defined on ; here, the elements (edges) in are interpreted as being deleted, while those in are contracted, so that the support of the resulting minor is given by .This is why our incidence algebra is highly useful in matroid theory: it can be employed just like formal power series, notably by supporting operations such as differentiation and substitution. This approach is closely linked to the axiomatic characterizations of matroids and to a careful study of the greedy algorithm. Consequently, it should enable us to tackle, at the very least, every exact enumeration problem related to matroids and oriented matroids (including Tutte polynomials). To illustrate this idea, we provide concise formulations and proofs for many classical and new results in the theories of graphs, matroids, and oriented matroids, emphasizing the benefits of studying a matroid together with its minors as a single object - what we call a matroid power series
A localisation phase transition for the catalytic branching random walk
No full textWe show the existence of a phase transition between a localisation and a non-localisation regime for a branching random walk with a catalyst at the origin. More precisely, we consider a continuous-time branching random walk that jumps at rate one, with simple random walk jumps on , and that branches (with binary branching) at rate everywhere, except at the origin, where it branches at rate . We show that, if is large enough, then the occupation measure of the branching random walk localises (i.e.~when normalised by the total number of particles, it converges almost surely without spatial renormalisation), whereas, if is close enough to , then the occupation measure delocalises, in the sense that the proportion of particles in any finite given set converges almost surely to zero. The case (when branching only occurs at the origin) has been extensively studied in the literature and a transition between localisation and non-localisation was also exhibited in this case. Interestingly, the transition that we observe, conjecture, and partially prove in this paper occurs at the same threshold as in the case~.One of the strengths of our result is that, in the localisation regime, we are able to prove convergence of the occupation measure, whilst existing results in the case give convergence of moments instead
A parametric study of the turbulent energy transfer in three-dimensional Lattice-Boltzmann Hall-MHD plasma simulations
No full textInternational audienceObservations from heliospheric missions have deepened our understanding of the solar wind. However, the complexity arising from its quasi-collisionless and multiscale nature leaves open several key questions on how energy is transferred and dissipated in the interplanetary plasma. In this parametric study, we investigate turbulent energy transfer and intermittent plasma processes in three-dimensional direct numerical simulations, in the Hall-MHD framework, performed with the Fast Lattice-Boltzmann Algorithm for Magnetohydrodynamics Experiments (FLAME) . We analyze properties of the global and pseudo-local energy transfer, to characterize the scale-dependent behavior of the energy cascade in Hall-MHD plasmas by means of the so called local energy transfer proxy (or LET), the analysis of the probability distribution functions of the macroscopic fields, and the flatness of their increments. Our results indicate that scale-dependent structures, small-scale intermittency, and statistical properties in the plasma flow are sensitive to variations of the Hall term, providing insights that are potentially useful for the interpretation of space plasma dynamics
Alignement d'ontologies frugal pour servients WoT
No full textInternational audienceIn distributed and constrained Web of Things (WoT) infrastructures, servients exchange RDF messages that can rely on compact formats to reduce both bandwidth and memory footprint. To this end, in the context of the CoSWoT project, we adapted the CBOR-LD specification. As servients can connect or appear in these infrastructures, each using a different dialect, there is a need for off-the-shelf interoperability during message exchanges at runtime. This paper presents a frugal workflow to detect semantic similarity among messages encoded in CBOR-LD using heterogeneous vocabularies. We experimented different ML-based approaches: a lightweight neural network, a decision tree-based model and a transformer-based architecture. We present a end-to-end pipeline able to integrate each of these algorithms and provide alignment predictions. We ran our experiments on a catalogue of more than 700 production payloads curated to train the models. Our results show various levels of prediction accuracy regarding approaches and model sizes, leading to appropriate choices depending on available resources on targeted platforms.Dans une infrastructure Web des Objets (WoT) distribuée et contrainte, des servients échangent des messages RDF pouvant s’appuyer sur des formats compacts afin de réduire à la fois la bande passante et l’empreinte mémoire. À cette fin, dans le cadre du projet CoSWoT, nous avons adapté la spécification CBOR-LD. Comme les servients peuvent se connecter ou apparaître au sein de ces infrastructures, chacun utilisant un dialecte différent, il est nécessaire d’assurer à l'exécution et de manière automatique une interopérabilité entre les servients à partir des messages qu'ils échangent. Cet article présente une méthodologie permettant de détecter de manière frugale la similarité sémantique entre des messages encodés en CBOR-LD utilisant des vocabulaires hétérogènes. Nous avons expérimenté différentes approches basée sur l’apprentissage automatique : un réseau de neurones léger, un modèle basé sur des arbres de décision et une architecture transformeur. Nous présentons un pipeline capable d’intégrer chacun de ces algorithmes et de fournir des prédictions d’alignement. Nous avons mené nos expériences sur un catalogue de plus de 700 messages produits par des servients, sélectionnés pour entraîner les modèles. Nos résultats montrent divers niveaux de précision des prédictions selon les approches et la taille des modèles, permettant de choisir une approche en fonction des ressources disponibles sur les plateformes visées
RibPull: Implicit Occupancy Fields and Medial Axis Extraction for CT Ribcage Scans
No full textInternational audienceWe present RibPull, a methodology that utilizes implicit occupancy fields to bridge computational geometry and medical imaging. Implicit 3D representations use continuous functions that handle sparse and noisy data more effectively than discrete methods. While voxel grids are standard for medical imaging, they suffer from resolution limitations, topological information loss, and inefficient handling of sparsity. Coordinate functions preserve complex geometrical information and represent a better solution for sparse data representation, while allowing for further morphological operations. Implicit scene representations enable neural networks to encode entire 3D scenes within their weights. The result is a continuous function that can implicitly compesate for sparse signals and infer further information about the 3D scene by passing any combination of 3D coordinates as input to the model. In this work, we use neural occupancy fields that predict whether a 3D point lies inside or outside an object to represent CT-scanned ribcages. We also apply a Laplacian-based contraction to extract the medial axis of the ribcage, thus demonstrating a geometrical operation that benefits greatly from continuous coordinate-based 3D scene representations versus voxel-based representations. We evaluate our methodology on 20 medical scans from the RibSeg dataset, which is itself an extension of the RibFrac dataset. We will release our code upon publication.</div