INRIA a CCSD electronic archive server
Not a member yet
122212 research outputs found
Sort by
A cloud-based platform for seamless digital testing of cardiac medical devices and drugs
No full textInternational audienceA cloud-based platform for seamless digital testing of cardiac medical devices and drugs Cardiovascular diseases affect 15 million people in Europe, and digital solutions are now seen as very useful tools in the search for new drugs and medical devices. SimCardioTest is a four-year project funded by the European Commission that aims to develop credible computer modelling and simulation approaches on a cloud-based platform for testing cardiac drugs and devices in silico
Special Sessions -Emerging Scope and Design Challenges for Approximate Computing: Optimizing Accuracy-PPA trade-offs and Beyond
No full textInternational audienceThe rapid growth of AI workloads is driving interest in Approximate Computing (AxC) as a means to enable low-cost, energy-efficient inference in resource-constrained systems. By introducing controlled inaccuracies, AxC can deliver substantial gains in power, performance, and area (PPA) while leveraging the inherent error tolerance of many AI models. Achieving this potential requires adapting existing frameworks to support the design and optimization of neural networks with approximate operators. Modern AxC research extends beyond accuracy-PPA trade-offs to address reliability and security, reducing redundancy overheads and exploring the distinctive side-channel implications of approximation. Application-aware approaches, such as those for spiking neural networks, show that tailoring approximation to workload-specific error behavior can surpass generic strategies. This article examines AI-guided design methods and the interplay between efficiency, reliability, and security, highlighting how these interconnected facets can advance embedded and high-performance computing.</div
On the non-convexity issue in the radial Calderón problem
No full textA classical approach to the Calderón problem is to estimate the unknown conductivity by solving a nonlinear least-squares problem. It leads to a nonconvex optimization problem which is generally believed to be riddled with bad local minimums. We revisit this issue in the case of piecewise constant radial conductivities and prove that, contrary to previous claims, there are no spurious critical points in the case of two scalar unknowns with no measurement noise. We also provide a partial proof of this result in the general setting which holds under a numerically verifiable assumption. Finally, we investigate whether a recently proposed approach based on convexification yields better reconstructions. For the first time, we propose a way to implement it in practice and show that it is consistently outperformed by some least squares solvers, which are also faster and require less measurements
Reinfection induced multistability in an epidemic model
No full textInternational audienceWe consider in this paper the effects induced on the dynamics of disease transmission by differences between primary and subsequent infections (i.e. reinfections), due e.g. to enhancement or weakening of the susceptibility or infectivity. To this end, an 8-dimensional 'two-stage' SEIRS reinfection model is considered, extending the classical 4-dimensional SEIRS model. We characterize the steady states of the model according to the basic reproduction number R0. It is shown that the reinfection induced heterogeneity may cause up to two endemic equilibria when R0 ď 1, and up to three endemic equilibria otherwise. Specifically, this suggests that the model may present backward bifurcation for R0 ă 1, a feature quite important from the point of view of disease control; and two successive saddle node bifurcations for R0 ą 1. Simulations confirm this situation. The disease persistence of the model is also examined. Finally, we prove, for two specific SIRI and SEIRE models accounting for partial immunity and demography, the asymptotic convergence of every trajectory to a steady state. Notably, these particular cases still admit up to two endemic equilibria (when R0 ă 1), which makes this result non trivial. The proof is based on an application of Li & Muldowney theory to epidemiological systems with multiple endemic equilibriums
Full-scale experimental and numerical analysis of pavement surface layers responses to different loading conditions
No full textInternational audienceWhile base pavement layers have been often studied due to their main contribution to fatigue resistance, less studies have focused on surface layers. This study investigates the mechanical response of surface layers under different loading conditions, including different axle configurations: single and tridem axles and different interface bonding conditions, using experimental and numerical approaches. Two fullscale pavements with different types of surface layer materials, were subjected to those loading configurations. Strain gauges and optical fibers embedded at different depths were used to analyze strain responses under varying surface layer temperatures. Results show that tridem axle generates higher strains, due to the interaction between the closely spaced wheels. In addition, one structure presented a well bonded surface layer, leading to compressive strains in the surface layer; on the other structure interface debonding developed tensile strains at the bottom of the surface layer. Numerical predictions using Viscoroute©2.0 captured the effects of axle configurations and interface conditions
A Generalized Potential Game Approach of UAV Swarm Coordination for Hidden Target Localization
No full textInternational audienceConsidering a swarm of Unmanned Aerial Vehicles (UAVs) carrying sensors with nondeterministic detection and noisy localization measurements, while sharing observations with neighboring UAVs, we address the problem of localization of a hidden target in continuous space and discrete time. The goal is to coordinate UAVs to maximize the information gathered while minimizing their individual energy costs. We formulate the problem as a time-varying non-cooperative game with coupling constraints. We show that the target is localized in finite time with probability one and that the game has a generalized potential structure. Further, we provide an exact best-response algorithm for UAVs to iteratively compute their trajectories. Finally, we numerically compare the potential game to the team-based approach, demonstrating comparable performance under different communication graph structures and assessing the impact of the swarm size on various metrics
Méthodes computationnelles et analyse des réseaux temporels : applications en neurosciences
No full textThis thesis develops computational methods for the analysis of temporal networks, with an emphasis on applications to neuroscience. Temporal graphs provide a natural representation for dynamic systems in which interactions evolve over time, such as functional connectivity in the human brain. As a relatively recent area of study, temporal graph theory still requires the development of dedicated tools and generalizations of classical static methods. To address this gap, new metrics are proposed to extend classical notions, such as clustering coefficient, path length, and small-worldness, into the temporal domain. These are designed to capture both local and global structure while respecting the causality imposed by time. A novel null model, the Random Temporal Hyperbolic model, is introduced to simulate the small-world property observed in brain dynamics while randomizing other topological features. This provides a meaningful baseline for the statistical evaluation of temporal connectivity patterns. A machine learning framework is also introduced to identify task-relevant subnetworks from temporal connectivity data. This approach integrates Shapley values to quantify the contribution of each subnetwork to the model's predictions. Applied to fMRI recordings during naturalistic stimuli, the framework reveals interpretable patterns in how different brain regions support narrative comprehension, aligning with current cognitive neuroscience theories. To support and generalize these computational approaches, I contributed to developed GraphNeuralNetworks.jl, an open-source Julia package for building and training graph neural networks on static, temporal, and heterogeneous graphs. This flexible and performant software enables rapid experimentation with modern graph-based learning models and is used here in the context of both brain decoding and traffic forecasting. Altogether, this thesis introduces a collection of reusable mathematical, algorithmic, and software tools designed for modeling and analyzing temporal graphs, while contributing to our understanding of the brain's dynamic architecture. Though they were initially developed in neuroscience, these methods can be applied to a broad range of disciplines that deal with evolving relational data.Cette thèse développe des méthodes computationnelles pour l'analyse des réseaux temporels, avec un accent particulier sur les applications en neurosciences. Les graphes temporels offrent une représentation naturelle des systèmes dynamiques dans lesquels les interactions évoluent au fil du temps, comme la connectivité fonctionnelle dans le cerveau humain. En tant que domaine d'étude relativement récent, la théorie des graphes temporels nécessite encore le développement d'outils dédiés et la généralisation de méthodes classiques issues du cadre statique. Pour combler cette lacune, de nouvelles métriques sont proposées afin d'étendre des notions classiques, telles que le coefficient de clustering, la longueur des chemins et le caractère petit-monde, au domaine temporel. Ces métriques sont conçues pour capturer à la fois la structure locale et globale tout en respectant la causalité imposée par le temps. Un nouveau modèle nul, appelé modèle hyperbolique temporel aléatoire, est introduit afin de simuler la propriété de petit-monde observée dans les dynamiques cérébrales tout en rendant aléatoire d'autres caractéristiques topologiques. Ce modèle fournit une base pertinente pour l'évaluation statistique des motifs de connectivité temporelle. Un cadre d'apprentissage automatique est également introduit pour identifier les sous-réseaux pertinents à partir des données de connectivité temporelle. Cette approche intègre les valeurs de Shapley pour quantifier la contribution de chaque sous-réseau aux prédictions du modèle. Appliqué à des mesures réalisées par IRMf lors de stimuli naturels, ce cadre révèle des motifs interprétables illustrant comment différentes régions cérébrales soutiennent la compréhension narrative, en accord avec les théories actuelles en neurosciences cognitives. Pour soutenir et généraliser ces approches computationnelles, j'ai contribué au développement de GraphNeuralNetworks.jl, une bibliothèque open-source en Julia dédiée à la construction et à l'entraînement de réseaux neuronaux sur graphes statiques, temporels et hétérogènes. Ce logiciel flexible et performant permet des expérimentations rapides avec les modèles modernes d'apprentissage sur graphes, et est utilisé ici dans le contexte du décodage cérébral et de la prévision du trafic. Dans l'ensemble, cette thèse introduit un ensemble d'outils mathématiques, algorithmiques et logiciels réutilisables, conçus pour modéliser et analyser les graphes temporels, tout en contribuant à notre compréhension de l'architecture dynamique du cerveau. Bien que développées initialement pour les neurosciences, ces méthodes peuvent être appliquées à un large éventail de disciplines traitant de données relationnelles évolutives
Contrasting glucosinolate profiles in rapeseed genotypes shape the rhizosphere-insect continuum and microbial detoxification potential in a root herbivore
No full textInternational audiencePlant secondary metabolites are key mediators of plant-insect-microbiome interactions, yet their role in structuring functionally relevant insect-associated microbial communities remains poorly understood. Here, we combined a factorial experiment using Brassica napus genotypes differing in glucosinolate (GLS) content with distinct succession to investigate the eco-evolutionary dynamics of the microbiota of the root herbivore Delia radicum. Amplicon sequencing and microbial culturing revealed that both rhizospheric and gut microbial communities are shaped by plant genotype and soil legacy, with a subset of bacterial taxa shared across compartments. Notably, Pseudomonas brassicacearum, harboring the isothiocyanates (ITC) detoxifying gene saxA, was consistently recovered from both plant and insect habitats. Functional assays confirmed its capacity to degrade 2-phenylethyl isothiocyanate (PEITC), a major toxic GLS hydrolysis product. Other gut-derived microbial isolates exhibited heterogeneous responses to PEITC, ranging from growth inhibition, promotion, or growth recovery after a prolonged lag phase. Despite the toxicity of ITC, insect fitness proxies were enhanced on GLS +plants, suggesting microbiota-mediated adaptation to host chemical defenses. Our findings reveal a plant genotype-specific filtering of environmentally acquired microbes and highlight the role of detoxifying symbionts in Delia radicum performance
Classifying mental motor tasks from chronic ECoG-BCI recordings using phase-amplitude coupling features
No full textInternational audienceIntroduction: Phase-amplitude coupling (PAC), the modulation of highfrequency neural oscillations by the phase of slower oscillations, is increasingly recognized as a marker of goal-directed motor behavior. Despite this interest, its specific role and potential value in decoding attempted motor movements remain unclear.Methods: This study investigates whether PAC-derived features can be leveraged to classify different motor behaviors from ECoG signals within Brain-Computer Interface (BCI) systems. ECoG data were collected using the WIMAGINE implant during BCI experiments with a tetraplegic patient performing mental motor tasks. The data underwent preprocessing to extract complex neural oscillation features (amplitude, phase) through spectral decomposition techniques. These features were then used to quantify PAC by calculating different coupling indices. PAC metrics served as input features in a machine learning pipeline to evaluate their effectiveness in predicting mental tasks (idle state, right-hand movement, left-hand movement) in both o ine and pseudo-online modes. Results:The PAC features demonstrated high accuracy in distinguishing among motor tasks, with key classification features highlighting the coupling of theta/low-gamma and beta/high-gamma frequency bands.Discussion: These preliminary findings hold significant potential for advancing our understanding of motor behavior and for developing optimized BCI systems.</div
Formalization of Brownian motion in Lean
No full textBrownian motion is a building block in modern probability theory. In this paper, we describe a formalization of Brownian motion using the Lean theorem prover. We build on the existing measure-theoretic foundations in Lean's mathematical library, Mathlib, and we develop several key components needed for the construction of Brownian motion, including the Carathéodory and Kolmogorov extension theorems, Gaussian measures in Banach spaces, and the Kolmogorov-Chentsov theorem for path continuity