Archive ouverte de Centrale Lyon
Not a member yet
32420 research outputs found
Sort by
Applying Operational Modal Analysis to rammed earth buildings for seismic assessment: an exploratory study
No full textInternational audienceApplying Operational Modal Analysis to rammed earth buildings for seismic assessment: an exploratory stud
A reliable activity proxy in SPIRou spectra of M dwarfs using machine learning
No full textInternational audienceContext. Recent instruments have extended radial velocity observations from the optical domain to the near-infrared range (NIR). In particular, this has allowed M dwarfs to be studied more extensively, which is notable because they are known to host rocky planets more frequently. However, these stars also have, on average, stronger magnetic activity compared to solar-type stars, and investigating this magnetic activity is key to uncovering any planets around such stars.Aims. This paper aims to extensively test a new reliable magnetic activity indicator named W1 and confirm it as a proxy for the small-scale magnetic field for M dwarf stars.Methods. The magnetic activity indicator W1 is derived from a principal component analysis (PCA) applied on the per-line differential line width (dLW). However, the PCA is highly sensitive to contamination from telluric residuals in the spectra. Therefore, we employed a filtering technique based on such machine learning tools as the unsupervised dimensional reduction (DR) algorithm and support vector machine (SVM) to remove faulty lines. We assessed the performance of this filtering method using a simulation of observations of the per-line dLW variations before applying it to NIR high-resolution spectroscopic observations from SPIRou (at the Canada-France-Hawaii Telescope) of five targets with various stellar magnetic activity levels, spectral types, and rotation periods contained in the SPIRou Legacy Survey, namely AU Mic, EV Lac, GJ1286, GJ1289, and G1 410.Results. The filtered W1 signal is modulated with a period consistent with the rotation period retrieved from activity indicators, corresponding to the magnetic activity for all the stars studied. It also correlates with the small-scale magnetic field of all five stars, with a direct Pearson correlation coefficient greater than 0.80. Additionally, we identified 201 stellar lines that are particularly sensitive to magnetic activity that could be valuable for the study of magnetic fields
Non-Euclidean Geometries for Prototype-Based Image Classification: a Reality Check
No full textInternational audienceIdentifying the most appropriate metric to capture similarities between embeddings remains a longstanding challenge in Machine Learning (ML). Recent research has highlighted the potential of non-Euclidean geometries for modeling embedding distances, particularly in datasets with latent hierarchical structures. However, selecting an optimal geometry is non-trivial, and the literature lacks a comprehensive analysis of the advantages and limitations associated with different metric spaces. In this paper, we aim to address this gap by focusing on a setting where the choice of geometry plays a crucial role in the optimization process: Prototype Learning (PL). Through extensive comparisons across a diverse range of datasets, we uncover key insights into the behavior of non-Euclidean spaces, showcasing their limitation and possible future developments
A SEMI-PARAMETRIC ESTIMATION FOR MAX-MIXTURE SPATIAL PROCESSES
No full textInternational audienceWe proposed a semi-parametric estimation procedure in order to estimate the parameters of a max-mixture model and also of a max-stable model (inverse max-stable model) as an alternative to composite likelihood. A good estimation by the proposed estimator required the dependence measure to detect all dependence structures in the model, especially when dealing with the max-mixture model. We overcame this challenge by using the F-madogram. The semi-parametric estimation was then based on a quasi least square method, by minimizing the square difference between the theoretical F-madogram and an empirical one. We evaluated the performance of this estimator through a simulation study. It was shown that on an average, the estimation is performed well, although in some cases, it encountered some difficulties. We apply our estimation procedure to model the daily rainfalls over the East Australia
Cubical confluences and formalisation of cubical categories
No full textThis thesis belongs to a research programme on the formalisation of rewriting. We study the confluence property, which guarantees that two rewrites on the same expression can be extended by other rewriting paths leading to a common expression. This fundamental property of rewriting has been expressed in various languages: diagrams, inclusions of relations, Kleene algebras, globular and cubical categories. This thesis studies the formalisation of this property from a cubical point of view. The confluence schemes of branchings of two rewrites are by definition square and those of k-branchings form k-cubes. This suggests a natural formulation of confluence proofs in the language of cubical categories and polygraphs. In this thesis, we introduce a setting for studying confluence proofs as cubical constructions and deriving single-set cubical formalisations. We define the notion of cubical contraction which allows us to prove the acyclicity of cubical (ω,0)-categories. We show how to construct cubical contractions from confluent and terminating abstract rewriting systems. We then give an explicit construction of a cubical polygraphic resolution consisting in every dimension of the cubical shapes of the confluences of the multiple branchings. We formulate confluence proof schemes à la Newman and Church-Rosser in this polygraphic language. We introduce single-set cubical categories, which axiomatise cubical categories by defining the dimension of cells by fixed-point equations. The single-set axiomatisation is well suited to a formalisation in the proof assistant Isabelle/HOL. We formalise single-set cubical ω-categories in Isabelle/HOL. Finally, we formalise the notion of rewriting system in polygraphable single-set globular ω-categories. They correspond to the globular ω-categories freely generated by ω-polygraphs. We characterise them as the cofibrant objects of the transferred folk model structure and formalise them in Isabelle/HOL.Cette thèse s'inscrit dans un programme de recherche sur la formalisation de la réécriture. Nous étudions la propriété de confluence, qui garantit que deux réécritures sur une même expression peuvent être prolongées par d'autres chemins de réécriture conduisant à une expression commune. Cette propriété fondamentale de la réécriture a été exprimée dans différents langages: diagrammes, inclusions de relations, algèbres de Kleene, catégories globulaires et cubiques. Cette thèse étudie la formalisation de cette propriété d'un point de vue cubique. Les schémas de confluence de branchements de deux réécritures sont par définition carrés et ceux de k-branchements forment des k-cubes. Cela suggère une formulation naturelle des preuves de confluence dans le langage des catégories et polygraphes cubiques. Dans cette thèse, nous introduisons un cadre permettant d'étudier les preuves de confluence comme des constructions cubiques et d'en déduire des formalisations cubiques à un seul ensemble. Nous explicitons la notion de contraction cubique permettant de démontrer l'acyclicité des (ω,0)-catégories cubiques. Nous montrons comment construire des contractions cubiques à partir de systèmes de réécriture abstraits confluents et terminants. Nous donnons ainsi une construction explicite de résolution polygraphique cubique constituée en toute dimension par les formes cubiques des confluences des multiples branchements. Nous formulons les schémas de preuves de confluence à la Newman et Church-Rosser dans ce langage polygraphique. Nous introduisons les catégories cubiques à un seul ensemble, dites single-set, qui axiomatisent les catégories cubiques en définissant la dimension des cellules par des équations de point fixe. L'axiomatisation à un seul ensemble est bien adaptée à une formalisation dans l'assistant de preuve Isabelle/HOL. Nous formalisons les ω-catégories cubiques à un seul ensemble dans Isabelle/HOL. Enfin, nous formalisons le concept de système de réécriture dans les ω-catégories globulaires polygraphiables à un seul ensemble, qui correspondent aux ω-catégories globulaires librement engendrées par des ω-polygraphes. Nous les caractérisons comme les objets cofibrants de la structure de modèle folk transférée et nous les formalisons dans Isabelle/HOL
Machine learning-based reduction and unbiased reinforcement learning method for QoS-aware services composition in IoT environments
No full textInternational audienceThe selection and composition of competitive functional services to satisfy the user’s evolving needs is an open challenge due to the increasing number of services offered by applications in internet of things (IoT) environments. This challenge can be handled using optimization methods based on machine learning (ML) and software agent models to find the most appropriate services in terms of quality of service (QoS) criteria values. Among these methods, several reinforcement learning (RL) techniques were exploited in the literature. However, most of them cause biased learning and consider the overall set of services during the composition process. Indeed, biased learning may lead the RL-based approaches to a convergence failure of the composition mechanism. This failure along with considering the overall set of services induce a performance degradation in terms of composition time and cumulative reward especially when the services space becomes larger. This paper proposes to combine machine learning-based reduction techniques with the Monte-Carlo method for services composition (ML-MCSC) in IoT environments. Unlike most existing services composition algorithms that consider the overall set of services, the ML-MCSC approach exploits two ML techniques (random forest and t-distributed stochastic neighbor embedding) to reduce the number of services that will be considered in the composition process. This reduction improves the QoS cumulative reward and decreases the composition time since the unpromising services in terms of QoS are filtered out. Furthermore, the services composition is modeled as a Markov decision process (MDP) and solved using the unbiased on-policy Monte Carlo method with adaptive decayed epsilon-greedy strategy. Simulation results obtained on WSC-2009 dataset show that the ML-MCSC approach performs better compared to five existing services composition approaches, in terms of computation time, composition’s QoS cumulative reward and success rate
Individual Color Matching Functions Estimated from Spectrally Narrow-Band Achromatic Matches Using Physiological Observer Models
No full textInternational audienc
Welcome to London Imaging Meeting 2025: Lighting and Imaging
No full textInternational audienceAbstract Following the tradition of previous year’s successful research events dedicated to the future of imaging science, the 6th London Imaging Meeting (LIM) takes place at the Institute of Physics (IoP) in London, UK, from September 8 to 10, 2025. The theme chosen for this year is “Lighting and Imaging” with the intention of bringing together students, researchers, engineers and others from these two distinct fields of science and technology, which share so many fundamental concepts and methods, and whose research challenges are often deeply intertwined. Just think about trying to capture good quality images in really low light levels! The first day features a summer school with four insightful short courses on radiometry, photometry and colour science for lighting; materials, lights, algorithms, and software for light transport simulation; the colourful math behind multichannel imaging systems; and methods for illuminant estimation and correction in colour imaging workflows. List of Conference Chairs, Series chairs and Steering Committee are available in this PDF
High pressure study of exotic hexagonal phase of Ge grown by molecular beam epitaxy on self-assisted GaAs nanowires
No full textNational audienc
Low losses optical devices on GaP/GaAs platform
No full textInternational audienceHere we study Gallium Phosphide-based devices made from GaP/GaAs epilayers in the framework of non-linear photonic integration. We demonstrate state-of-the art propagation losses in the near infrared and discuss this value in terms of roughness and crystal defects