Archive ouverte HAL-LAAS
Not a member yet
12189 research outputs found
Sort by
Transport and deformation in soft / porous tissues: mechanisms and opportunities
No full textInternational audienc
HIGH-THROUGHPUT MECHANOBIOLOGICAL CELL DISCRIMINATION USING AUTOMATED AFM AND MACHINE LEARNING
No full textInternational audienceMechanobiological measurements offer a promising avenue for distinguishing healthy cells from pathological ones. However, a major limitation of atomic force microscopy (AFM)—a widely used technique for such measurements—is its low throughput and lack of standardization[1]. In this study, we optimized AFM-based mechanical measurements on cell populations and developed a novel technology that integrates cell patterning with AFM automation, significantly increasing measurement efficiency. [2]Our system enables the acquisition of mechanical data from hundreds of cells, with 956 cells analyzed in this study. For each cell, 16 force curves (FCs) were recorded, and seven key mechanical features per FC were extracted, forming a comprehensive mechanome dataset. To classify these measurements, we employed a machine learning-based approach using a fuzzy logic algorithm trained to distinguish between nonmalignant and cancerous cells. The training dataset included up to 120 cells per cell line.As a proof of concept, we first applied our method to prostate cell lines—nonmalignant RWPE-1 and cancerous PC3-GFP—before extending it to skin fibroblast lines—nonmalignant Hs 895.Sk and cancerous Hs 895.T. Despite a high degree of similarity across measurements (ranging from 79% to 100%), our method achieved a classification accuracy of 73% on a validation dataset comprising 194 cells per cell line.These results demonstrate the potential of combining AFM automation with machine learning for high-throughput mechanobiological cell classification. This approach not only enhances measurement efficiency but also provides a standardized framework for analyzing cell mechanics, paving the way for future applications in cancer diagnostics and mechanobiology research.Références : exemple de format ci-dessous[1] Thomas- -Chemin, et al. ACS Nano 19.5 (2025): 5045-5062[2] Thomas - - Chemin, et al., ACS Applied Materials and Interfaces 16.34 (2024): 44505-44517Adresse mail : [email protected]
HIGH-THROUGHPUT MECHANOBIOLOGICAL CELL DISCRIMINATION USING AUTOMATED AFM AND MACHINE LEARNING
No full textInternational audienceMechanobiological measurements offer a promising avenue for distinguishing healthy cells from pathological ones. However, a major limitation of atomic force microscopy (AFM)—a widely used technique for such measurements—is its low throughput and lack of standardization[1]. In this study, we optimized AFM-based mechanical measurements on cell populations and developed a novel technology that integrates cell patterning with AFM automation, significantly increasing measurement efficiency. [2]Our system enables the acquisition of mechanical data from hundreds of cells, with 956 cells analyzed in this study. For each cell, 16 force curves (FCs) were recorded, and seven key mechanical features per FC were extracted, forming a comprehensive mechanome dataset. To classify these measurements, we employed a machine learning-based approach using a fuzzy logic algorithm trained to distinguish between nonmalignant and cancerous cells. The training dataset included up to 120 cells per cell line.As a proof of concept, we first applied our method to prostate cell lines—nonmalignant RWPE-1 and cancerous PC3-GFP—before extending it to skin fibroblast lines—nonmalignant Hs 895.Sk and cancerous Hs 895.T. Despite a high degree of similarity across measurements (ranging from 79% to 100%), our method achieved a classification accuracy of 73% on a validation dataset comprising 194 cells per cell line.These results demonstrate the potential of combining AFM automation with machine learning for high-throughput mechanobiological cell classification. This approach not only enhances measurement efficiency but also provides a standardized framework for analyzing cell mechanics, paving the way for future applications in cancer diagnostics and mechanobiology research.Références : exemple de format ci-dessous[1] Thomas- -Chemin, et al. ACS Nano 19.5 (2025): 5045-5062[2] Thomas - - Chemin, et al., ACS Applied Materials and Interfaces 16.34 (2024): 44505-44517Adresse mail : [email protected]
Peak Time-Windowed Risk Estimation of Stochastic Processes
No full textThis paper develops a method to upper-bound extreme-values of time-windowed risks for stochastic processes. Examples of such risks include the maximum average or 90% quantile of the current along a transmission line in any 5-minute window. This work casts the time-windowed risk analysis problem as an infinite-dimensional linear program in occupation measures. In particular, we employ the coherent risk measures of the mean and the expected shortfall (conditional value at risk) to define the maximal time-windowed risk along trajectories. The infinite-dimensional linear program must then be truncated into finite-dimensional optimization problems, such as by using the moment-sum of squares hierarchy of semidefinite programs. The infinite-dimensional linear program will have the same optimal value as the original nonconvex risk estimation task under compactness and regularity assumptions, and the sequence of semidefinite programs will converge to the true value under additional properties of algebraic characterization. The scheme is demonstrated for risk analysis of example stochastic processes
Data-Driven Azimuthal RHEED Construction for In Situ Crystal Growth Characterization
No full textInternational audienceReflection High-Energy Electron Diffraction (RHEED) is a powerful tool to probe the surface reconstruction during MBE growth. However, raw RHEED patterns are difficult to interpret, especially when the wafer is rotating. A more accessible representation of the information is therefore the so-called Azimuthal RHEED (ARHEED), an angularly resolved plot of the electron diffraction pattern during a full wafer rotation. However, ARHEED requires precise information about the rotation angle as well as of the position of the specular spot of the electron beam. We present a Deep Learning technique to automatically construct the Azimuthal RHEED from bare RHEED images, requiring no further measurement equipment. We use two artificial neural networks: an image segmentation model to track the center of the specular spot and a regression model to determine the orientation of the crystal with respect to the incident electron beam of the RHEED system. Our technique enables accurate, and potentially real-time ARHEED construction on any growth chamber equipped with a RHEED system
Slotted Reinforcement Learning-based radio resource allocation in sliced 5G networks
No full textInternational audience5G networks are designed for providing several different services as high speed Internet, low latency and M2M (Machine to Machine) communications. For enforcing such guaranteed services, slicing techniques are of essential importance to ensure isolation between resources allocated to each of these services, especially at the level of RAN (Radio Access Networks) and its time/frequency matrix. Given the scarcity of radio resources, this paper aims at proposing efficient radio resource allocation algorithms and mechanisms for 5G networks, avoiding resource wastes and enforcing slices isolation. The proposed solution highlights a new way of using reinforcement learning, and more specifically the Double DQN (Deep Q-Network) algorithm, based on a slotted approach for 5G resource allocations. The slotted use of Double DQN evaluation exhibits its benefits in terms of allocation performance and low latency
Design of a 2D metallic photonic crystal for spectral control in thermophotovoltaic devices
No full textInternational audienceThermophotovoltaic (TPV) devices convert thermal radiation from a high-temperature emitter into electricity using a photovoltaic (PV) cell. To maximize power output and efficiency, optical and thermal management is crucial [1]. One approach involves using a selective emitter engineered to emit photons primarily with energies above the PV cell’s bandgap. While effective, such emitters often face thermal stability issues at high operating temperatures. An alternative strategy employs a blackbody emitter and spectrally selective optical filter placed above the PV cell to reflect unwanted photons back to the emitter. This filter must exhibit high transmittance for in-band photons (energy > bandgap) and high reflectance for out-of-band photons near the bandgap (energy < bandgap), while maintaining negligible absorption across the spectrum.Although multilayer structures have been widely used to achieve such spectral selectivity, they typically require many layers and exhibit sensitivity to the angle of incidence, limiting their broadband and omnidirectional performance. As a promising alternative, two-dimensional (2D) photonic crystals (PhCs) consisting of a periodic array of cylindrical holes on a host matrix offer tunable radiative properties. Figure 1 illustrates the TPV design incorporating a 2D metallic PhC inspired by a recent simulation work [2].In this work, we design and simulate a metallic 2D PhC using FDTD to achieve the desired spectral selectivity. The influence of key geometric parameters, such as the radius of the cylindrical holes, their periodicity and thickness, on the PhC’s optical response is investigated. Additionally, we analyze the angular stability of the design.[1] B. Roux et al., Journal of Photonics for Energy, 14(4):042403–1, 2024.[2] S. Zhang et al., Optics Express, 31(6):9186-9195, 2023
Dynamic time series segmentation for health monitoring of hybrid systems
No full textInternational audienceMonitoring and diagnosing complex, real-world, industrial hybrid systems require accurate and up-to-date models that can adapt to evolving system behaviors. Such systems, characterized by both continuous and discrete dynamics, are best represented by hybrid models. In this article, we present the segmentation step of HyMED (Hybrid Model Enrichment for Diagnosis), a model-based health monitoring and diagnosis method that monitors hybrid systems and automatically updates the system model if necessary. HyMED uses noisy multivariate time series data to dynamically update models, addressing unanticipated degradations and faults. A key feature of HyMED is its online and passive segmentation step (ODS), which enables robust detection of system mode changes in complex, nonlinear time series. Unlike traditional segmentation methods, ODS dynamically determines its segmentation hyperparameters through an automatic parameter selection process. ODS guarantees adaptability without the need for manual adjustment. The effectiveness of HyMED's segmentation method is demonstrated through a case study on an engine timing system, where its performances are compared to the offline method depicted in the Ruptures library
Une méthode de planification basée sur les flux pour la progression multi-agents avec des agents déployables et des contraintes de communication
No full textInternational audienceThis paper deals with the problem of planning multiple agent movements through a mission area modeled as a graph. The agents undergo classic communication and temporal constraints, and the quantitative objective is the minimization of the team’s traversal makespan. Additional specificities make the problem a particularly complex routing one: on some nodes are associated durative and coordinated actions to perform, which can involve either the co-presence of several agents or time dependencies. Also, some agents are deployable and able to move on denser graphs: namely, aerial robots can take off and land on the ground vehicle at any planned position, and can fly above ground obstacles. We model the problem as a CSP and solve it with a network flow model. Results show the efficacy of the model and resolution scheme, which provides solutions with one or two orders of magnitude smaller time than a numerical temporal hierarchical planning model, with only a few percent loss of optimality.Cet article traite du problème de planification de mouvements de multiples agents à travers une zone de mission modélisée comme un graphe. Les agents sont soumis à des contraintes classiques de communication et temporelles, et l'objectif quantitatif est la minimisation du temps d'exécution total de l'équipe. Des spécificités supplémentaires rendent le problème de routage particulièrement complexe : sur certains nœuds sont associées des actions duratives et coordonnées à effectuer, qui peuvent impliquer soit la co-présence de plusieurs agents soit des dépendances temporelles. De plus, certains agents sont déployables et capables de se déplacer sur des graphes plus denses : à savoir, les robots aériens peuvent décoller et atterrir sur le véhicule terrestre à toute position planifiée, et peuvent voler au-dessus des obstacles au sol. Nous modélisons le problème comme un CSP et le résolvons avec un modèle de flot de réseau. Les résultats montrent l'efficacité du modèle et du schéma de résolution, qui fournit des solutions avec un ou deux ordres de grandeur de temps en moins qu'un modèle de planification hiérarchique temporelle numérique, avec seulement quelques pourcents de perte d'optimalité
Numerical modeling and experimental study of self-propagating flame fronts in Al/CuO thermite reactions
No full textInternational audienceNanothermites are promising energetic materials as their high-temperature reaction driven by the oxidation of a metallic fuel associated with the reduction of an oxidizer, can exhibit extremely fast burning rates, exceeding hundreds of m s -1 . In addition, by modifying reactant size, stoichiometry and compaction conditions, reaction properties (temperature, intermediate reactions, by-products) and combustion rates can be tailored, making it possible to customize combustion properties for each application. Unfortunately, in spite of three decades of research in the field of thermites, there is no predictive physical models able to provide design guidelines to experimentalists. The reason of this is that the complex multiphasic physics governing thermite combustion, where combustion gases interact with burning particles, is still poorly understood and documented, while being the key step to depict the dynamics of the flame front. The purpose of this work is to propose a first one-dimensional (1D) model that describes the dynamics of the reaction front propagation in Al/CuO powdered thermite considering the reacting flow combined with heat transfer, chemistry and fluid flow. CuO was chosen as it is the widest used metallic oxidizer, that decomposes below the flame temperature, leading to a gas phase driven reaction. Separate mass, momentum and energy transport equations for the three phases, namely Al, CuO particles and gas mixture, are written in the frame of an Euler-Euler approach for multiphase reactive flows. These equations are coupled by modeled interphase transfer terms. The theoretical formulation and numerical methods are detailed. After validating the model with experimental case studies -specifically, the combustion of Al/CuO powder in open glass tubes -numerical experiments are performed to demonstrate the utility of the code in (i) analyzing the multiphase flow dynamics at the thermite flame front, and (ii) examining the critical powder characteristics that affect the burn rate