HAL-CentraleSupelec
Not a member yet
77624 research outputs found
Sort by
Actes de la 9ème édition du colloque Pédagogie et Formation du Groupe INSA: Adaptation des parcours de formation aux nouveaux enjeux des métiers de l'ingénieur : quelles transformations et quels impacts ?
No full textNational audienceL’évolution du métier d’ingénieur implique une nécessaire réévaluation des parcours de formation, deleurs contenus, des méthodes et outils d’enseignement-apprentissage. À partir de retours d’expérienceet d’expérimentations dans nos établissements, nous pourrons analyser l’impact de ces transformationssur les acquis des étudiants, le développement de leurs compétences et réfléchir aux moyens de favoriserleur engagement dans ces différents dispositifs.La 9e édition du colloque Pédagogie et Formation du Groupe INSA entend questionner ces problématiqueset permettre d’échanger sur les actions en place, le déploiement de projets à l’échelle du groupe (notamment INSA 2025) et les partenariats développés ces deux dernières années
Exploring Functionalized Gold Nanoparticles Using a Fine-tuned MACE Machine Learning Interatomic Potential
No full textInternational audienc
SCHC Header Compression for GeoNetworking protocol: Adaptation and Performance Evaluation
No full textInternational audienceThe growing number of autonomous vehicles (AVs) significantly increases wireless network load, mainly due to the frequent exchange of safety messages such as Cooperative Awareness Messages (CAMs). A common strategy to mitigate this congestion is to compress application-layer data in these messages. However, a detailed analysis indicates that the GeoNetworking protocol, used for routing in ITS-G5 networks, represents a substantial part of the message overhead (about 40% of the size of an unsigned CAM). Therefore, we propose compressing this protocol using a generic header compression framework-Static Context Header Compression (SCHC). Our work adapts SCHC framework to the GeoNetworking protocol by defining two distinct compression rules. One of these rules is generic, provides lower compression, and can be pre-installed in all SCHCsupported vehicles. The second rule aggressively compresses the packet by leveraging the local conditions and can be optimised for a group of AVs, providing additional compression. In addition, we assess the effects of SCHC compression on an IEEE 802.11p based ITS-G5 network by estimating the Channel Busy Ratio (CBR) and Packet Delivery Ratio (PDR) using analytical models and measuring the time on-air of SCHC compressed CAMs. Our findings indicate that by reducing time on-air by 13%, the compressed CAMs can decrease CBR by up to 19% and increase PDR by up to 6%. Moreover, they incur an additional overhead of approximately 5% for compression and 2% for decompression on the YoGoKo on-board unit
A generalised virtual chemistry formalism with standardised thermodynamics, kinetics and complex transport properties
No full textInternational audienceAccurately and efficiently modeling combustion remains a central challenge in the transition to low-carbon energy systems. Virtual chemistry has emerged as a powerful chemistry reduction strategy, relying on optimised species and reactions to reproduce reference flames behaviours at a fraction of the computational cost. However, previous formulations depend on tabulated parameters, limiting their extrapolation capabilities and integration into standard reactive flow solvers. To address these limitations, a new standardised virtual chemistry formalism has been introduced, bridging virtual and detailed chemistry through chemical equilibrium between real and virtual species. The present work extends the framework to accurately capture laminar flame structures by jointly optimising the thermodynamic, kinetic, and transport properties. Special attention is given to validating and generalising virtual mechanism structures, with the development of a novel architecture dedicated to pyrolysing fuels and multi-component fuel surrogates. Theresulting highly-reduced optimised mechanisms adopt architectures compatible with CHEMKIN and CANTERA solvers.The methodology is applied to representative decarbonised and conventional fuels: Hydrogen, Dodecane, which is a key component of Sustainable Aviation Fuels (SAFs), a multi-component SAF surrogate and a Jet A multi-component surrogate. Particular emphasis is placed on reproducing flame structures across a wide range of combustion regimes, including premixed and non-premixed, adiabatic and non-adiabatic conditions. The results demonstrate the ability of the proposed virtual mechanisms to replicate reference flame structures with drastically reduced complexity, chemical stiffness and CPU time, paving the way for straightforward and efficient high-fidelity combustion simulations
Off the Hamster Wheel: Rethinking Dialogue Research through a Meta-Analysis of the ACL Anthology 2024
No full textInternational audienceIn this paper, we take a meta-review approach to investigate how conversation is currently studied in the field by analysing papers from the ACL Anthology 2024. We retrieved 407 papers, which represents about 6.1% of the papers published in the selected venues, and manually reviewed them to determine the conversational task addressed, the corpora used, and the evaluation methods employed. Our analysis leads to several observations. First, dialogue systems represent about half of the papers of the ACL Anthology 2024 while more formal and analytical approaches cover only 12%. Second, many papers provide lacking corpus descriptions, which shows a detachment from the data which becomes a simple tool instead of one of the pillars NLP/CL applications should be based on. Third, the evaluation methods, in particular when it comes to dialogue systems, often do not assess the interactional aspects of these systems or rely on assumptions not backed up from evidence of the dialogue research community. We argue that the field would benefit from a renewed focus on analysis and formal representation of conversation, a richer evaluation culture that includes interactional quality, and more systematic practices regarding the data presentation in papers.Cet article propose une méta-revue des articles de l'ACL Anthology 2024 portant sur le dialogue. À l’aide d’une méthode de filtrage basée sur les titres, nous avons identifié 407 articles, soit environ 6,1\% des publications des conférences sélectionnées. Ces articles ont été examinés manuellement afin d’identifier la tâche conversationnelle étudiée, les corpus utilisés et les méthodes d’évaluation employées. Notre analyse met en évidence plusieurs tendances. D’une part, les systèmes de dialogue représentent environ la moitié des articles, tandis que les approches plus formelles ou analytiques ne constituent qu’environ 12\%. D’autre part, de nombreux travaux fournissent des descriptions très limitées des corpus, ce qui suggère un certain éloignement vis-à-vis des données. Enfin, les méthodes d’évaluation — notamment pour les systèmes de dialogue — prennent rarement en compte les aspects interactionnels de la conversation. Nous plaidons ainsi pour davantage d’analyse formelle, des pratiques d’évaluation plus riches et une documentation des données plus systématique
Analogies et logiques de variations : une étude préliminaire (version étendue)
No full textUne logique de variations associée à une logique (dite alors « statique »), permet de représenter des relations binaires entre interprétations. Or, une relation analogique « a est à b ce que c est à d » peut souvent être appréhendée par deux relations binaires. La première relie a et b d'une part et c et d, d'autre part. La seconde relie a et c d'une part et b et d, d'autre part. C'est ce qui motive l'étude de cet article, lequel introduit une famille de relations quaternaires, dites de pré-analogie, entre formules d'une logique statique, utilisant des formules d'une logique de variations. Une pré-analogie vérifie deux des postulats sur les analogies (la symétrie et l'échange des moyens) mais pas nécessairement la réflexivité. Au-delà de la définition d'une telle famille de pré-analogies, dont certaines sont des analogies, quelques premiers résultats sont montrés, notamment dans un cadre propositionnel. Mots-clés analogie, logiques de variations Notons que b -a et d -c appartiennent à H mais pas nécessairement à G. L'analogie arithmétique A ari sur B coïncide avec l'analogie minimale sur B. En effet, une caractérisation de A ari est que, pour a, b, c, d ∈ B, a : b :: ari c : d ssi abcd est de la forme uuvv ou de la forme uvuv (u, v ∈ B). Une autre analogie sur les booléens est A SK , définie par Sheldon Klein (et rappelée dans [9]). Elle peut se définir par a : b :: SK c : d si |b -a| = |d -c| (pour a, b, c, d ∈ B). Une caractérisation de A SK est que, pour a, b, c, d ∈ B, a : b :: ari c : d ssi abcd est de la forme uuvv, de la forme uvuv ou de la forme uvvu (u, v ∈ B). Les logiques de variationsUne logique de variations est un formalisme s'appuyant sur une autre logique, appelée alors logique statique. La section 3.1 introduit des hypothèses sur les « logiques statiques ». La section 3.2 définit ce qu'est une logique des variations s'appuyant sur une logique statique. La section 3.3 décrit des opérateurs sur les logiques de variations.</div
Quantum Control and General Recursion beyond the Unitary Case
No full textCoherent control, aka quantum control, is a central concept in quantum computing that is attracting increasing attention from both the quantum foundations and quantum software communities. Defining coherent control in the presence of recursion and measurement has long been known to be a major challenge. In particular, no-go results have been established for standard semantical domains like completely positive maps. We address this problem by introducing the first quantum programming language with recursion that allows for the coherent control of arbitrary quantum operations. We equip this language with both an operational and a denotational semantics that we prove to be adequate. To design these semantics, we show that combining coherent control, recursion, and measurement crucially requires describing the evolution of subprograms in the absence of input. To address this, the operational semantics takes into account a default evolution branch, while the denotational semantics uses the concept of coherent quantum operation, based on vacuum extensions. We strengthen the validity of our approach by developing an observational equivalence: two programs are equivalent if their probability of termination is the same in any context. The denotational semantics is shown to be fully abstract with respect to this observational equivalence
Compressing a petabyte of k-mer matrices
No full textInternational audienceEfficiently querying petabyte-scale genomic data has the potential to revolutionize research in fields like medicine, ecology, and agronomy. The recent kmindex approach [1], which was applied to a complete snapshot of the Sequence Read Archive (SRA), made possible the creation of logan-search.org [2], a powerful search engine. However, the massive size of the index—about one petabyte—presents considerable challenges in terms of maintenance and usability.The index consists of binary matrices where rows represent datasets, columns represent k-mers (or their hash values), and cell values indicate k-mer presence (0 or 1). In this work, we propose a practical way to compress such matrices through data preprocessing and block compression. The preprocessing consists of swapping original columns with the aim of creating long runs of identical bits. This reduces matrix entropy, thus making them more compressible for general-purpose compressors. We show that a naive TSP heuristic, such as one based on the Nearest-Neighbor, is efficient for performing this task. The block compression consists of splitting matrices into blocks, compressing each block separately using Zstandard, concatenating compressed blocks, and encoding the starting position of blocks using the Elias-Fano succinct data structure [3,4]. The main objective is to achieve sensible compression ratios while minimising the impact on index creation and query times.The efficiency of the pre-processing depends on the type of original data. Therefore, we also propose a metric that can be used to estimate the potential benefits of this step before any changes are made to the matrix. We present experimental results for various types of indexed data. In the best cases, block compression reduces the size of the original index by a factor of two, while the pre-processing step achieves an additional 60% compression ratio. We also present results on the petabyte-sized index of logan-search [2], which enables to estimate the efficiency of our approach on a big index built from extremely diverse sources.We will discuss some alternative compression methods that could better compress k-mer matrices
Exposition : À la limite — Innover à la mesure du monde — 2055
No full textExposition scientifiqueThis exhibition takes us to the year 2055, to a planet where ecosystems and peoples are thriving. It highlights innovations that respect Earth’s limits and serve the well-being of all communities—human and non-human alike.And yet, today, an ecological and climate disaster is underway. Innovation—often reduced to technological innovation—is presented as a miracle solution, when in reality everything is connected: our lifestyles, the climate, geopolitics, the economic system, freshwater shortages, and more. Trying to solve one problem without thinking systemically doesn’t make sense.So how did we get there? In this fictional and “desirable” future, what does it mean to innovate? What questions does it raise for research, technology, and engineering? This journey through time opens up a new imagination—one that motivates and inspires optimism.Cette exposition nous projette en 2055, sur une planète Terre où les écosystèmes et les peuples se portent bien. Elle y met en lumière des innovations qui respectent les limites de la Terre, au service du bien-être de toutes les populations, humaines et non-humaines. Pourtant, aujourd’hui, un désastre écologique et climatique est en cours. L’innovation — au sens d’innovation technologique — est présentée comme la solution miracle, alors qu’en réalité tout est lié : nos modes de vie, le climat, la géopolitique, le système économique, les pénuries d’eau douce, etc. Innover pour résoudre un problème sans penser de manière systémique est un non-sens. Alors, comment avons-nous fait ? Dans cet avenir, fictif et « désirable », que veut dire innover ? Quelles questions se posent pour les métiers de la recherche, des techniques et de l’ingénierie ? Ce voyage temporel nous ouvre un nouvel imaginaire motivant et source d’optimisme
Haptixel: Encoding Data through Cutaneous Force-based Encountered-Type Fingertip Haptics
No full textInternational audienceData visualization benefits from non-visual cues to enable people to understand information by engaging with it through multimodality, yet most approaches rely on cumbersome technologies or large scale artifacts, making them difficult to adapt to dynamic or complex datasets. In this paper, we explore the use of cutaneous haptics as a lightweight quantitative channel for visualization tasks, allowing users to feel data and interact with it dynamically. We present Haptixel, an open-source DIY encountered-type wearable providing force-feedback on the users' fingertips' pulp. We propose an interaction framework illustrating how Haptixel can be used to complement visualization tasks through combinations of force levels and contact types. We evaluate our approach in a pixel-art-like VR user study (n=16) where pixels color/height are associated to forces as a univariate value mapping. Results show that participants can retrieve information with Haptixel, and significantly discriminate 3D-data with at least four levels of forces; suggesting that cutaneous force-feedback can function for quantitative distinctions in visualization tasks