INRIA a CCSD electronic archive server
Not a member yet
122212 research outputs found
Sort by
Toward Fluoroscopy Guided Robotic Needle Insertion for Radio Frequency Ablation
No full textInternational audienceThis article presents a fluoroscopy image‐based registration method along with a comprehensive protocol for robotic needle insertion in radiofrequency ablation (RFA) to treat liver cancer. The proposed method uses real‐time fluoroscopic images acquired from a C‐ARM system and integrates an inverse finite element (FE) simulation to compute robotic commands for accurate and adaptive needle steering. The registration procedure is fully automated and involves the injection of multiple radiopaque markers into the liver, enabling precise anatomical registration and targeted tumor localization. A key challenge addressed in this work is the integration of this image‐based registration with the inverse biomechanical simulation used to guide the robot during insertion. We describe how registration constraints can be mapped onto the surface of the biomechanical model to ensure consistent alignment between image data and robotic actuation. Designed to be adaptable to varying levels of radiologist expertise and applicable across a wide range of tumor locations, this method provides a robust and versatile solution for improving the accuracy and safety of minimally invasive liver cancer treatments
AdaptiFlow: An Extensible Framework for Event-Driven Autonomy in Cloud Microservices
No full textInternational audienceModern cloud architectures demand self-adaptive capabilities to manage dynamic operational conditions. Yet, existing solutions often impose centralized control models ill-suited to microservices decentralized nature. This paper presents AdaptiFlow, a framework that leverages well-established principles of autonomous computing to provide abstraction layers focused on the Monitor and Execute phases of the MAPE-K loop. By decoupling metrics collection and action execution from adaptation logic, AdaptiFlow enables microservices to evolve into autonomous elements through standardized interfaces, preserving their architectural independence while enabling system-wide adaptability. The framework introduces: (1) Metrics Collectors for unified infrastructure/business metric gathering, (2) Adaptation Actions as declarative actuators for runtime adjustments, and (3) a lightweight Event-Driven and rule-based mechanism for adaptation logic specification. Validation through the enhanced Adaptable TeaStore benchmark demonstrates practical implementation of three adaptation scenarios targeting three levels of autonomy self-healing (database recovery), self-protection (DDoS mitigation), and self-optimization (traffic management) with minimal code modification per service. Key innovations include a workflow for service instrumentation and evidence that decentralized adaptation can emerge from localized decisions without global coordination. The work bridges autonomic computing theory with cloud-native practice, providing both a conceptual framework and concrete tools for building resilient distributed systems. Future work includes integration with formal coordination models and application of adaptation techniques relying on AI agents for proactive adaptation to address complex adaptation scenarios
Nonlinear reduced-order modeling for time-domain electromagnetics
No full textInternational audienceThis work is concerned with the design of fully data-driven reduced-order models in the field of time-domain electromagnetics and nanophotonics. Numerical modeling plays a crucial role in revealing the behavior of light and matter interactions at the nanoscale, exploiting computational schemes, such as the Finite-Differences (FD) Time-Domain method and, as in our case, the Discontinuous Galerkin (DG) Time-Domain method. In this context, we study reduced-order modeling (ROM) as a consequence of the pressing need for fast surrogate model capable of handling physically and geometrically parameterized electromagnetic problems. Traditional ROM techniques like the Proper Orthogonal Decomposition (POD) and the Greedy algorithm have already been investigated in the literature, along with their inherent limitations in effectively capturing nonlinear phenomena. Here, we exploit a deep learning-based ROM strategy, the Graph Convolutional Autoencoder (GCA) method [1], serving as a nonlinear extension of POD compression, harnessing the power of Graph Neural Networks (GNNs) to retain geometric structures within unstructured meshes. We present some preliminary results for simple 2D benchmarks showing promising directions toward 3D problems involving complex geometries
Estimer l'impact carbone des activités numériques de l'Observatoire de Paris
No full textCe projet fait suite à la réponse de l'Observatoire de Paris à l'appel à projets CNRS Bas Carbone 2023. L'Observatoire de Paris désirait en effet effectuer un état des lieux de l'empreinte carbone de son utilisation du numérique, afin de prendre des mesures de réduction de cette dernière. Le CNRS a proposé à EcoInfo de répondre à cette demande. Nous remercions toutes les personnes de l'Observatoire qui ont participé à cette étude et nous ont aidé à collecter les données nécessaires à sa réalisation
Deep modelling of electric field distribution for clinical electroporation ablation therapies
No full textInternational audienceElectroporation ablation is a promising non-surgical and minimally invasive tumor ablation technique. The objective of this paper is to propose a deep learning strategy for fast and accurate 3D computation of the electric field distribution during the ablation procedure. The computation of the electrostatic potential is accelerated using a well-designed convolutional neural network, which serves to initialize the iterative solver for matrix inversion. The proposed approach, which combine deep learning and more standard numerical strategies, is compared with state-of-the-art numerical schemes, evaluating its performance in terms of error on the solution, residuals of the numerical scheme, and computation time. As a proof of concept, we demonstrate its potential on a clinical case of electroporation ablation of a liver tumor, showing that the proposed approach can be advantageously combined with clinical data to provide real-time insights into the effective electroporation ablation
GRANITE : a Byzantine-Resilient Dynamic Gossip Learning Framework
No full textGossip Learning (GL) is a decentralized learning paradigm where users iteratively exchange and aggregate models with a small set of neighboring peers. Recent GL approaches rely on dynamic communication graphs built and maintained using Random Peer Sampling (RPS) protocols. Thanks to graph dynamics, GL can achieve fast convergence even over extremely sparse topologies. However, the robustness of GL over dy- namic graphs to Byzantine (model poisoning) attacks remains unaddressed especially when Byzantine nodes attack the RPS protocol to scale up model poisoning. We address this issue by introducing GRANITE, a framework for robust learning over sparse, dynamic graphs in the presence of a fraction of Byzantine nodes. GRANITE relies on two key components (i) a History-aware Byzantine-resilient Peer Sampling protocol (HaPS), which tracks previously encountered identifiers to reduce adversarial influence over time, and (ii) an Adaptive Probabilistic Threshold (APT), which leverages an estimate of Byzantine presence to set aggregation thresholds with formal guarantees. Empirical results confirm that GRANITE maintains convergence with up to 30% Byzantine nodes, improves learning speed via adaptive filtering of poisoned models and obtains these results in up to 9 times sparser graphs than dictated by current theory
ELA: Secure, Lightweight, and Zero-Touch Enrollment for IoT Devices
No full textInternational audienceWhen deploying large numbers of IoT devices, an enrollment protocol takes care of admitting each device into their target network for the first time. The protocol must be secure to block malicious actors, easy to operate to reduce cost, and lightweight due to bandwidth constraints. Solutions in literature either involve use of pre-shared keys, require perdevice user input, or have been designed for non-constrained environments. This paper introduces EDHOC with Lightweight Authorization (ELA), a protocol for securely authorizing enrollment of devices in constrained networks with support for zerotouch deployments. We define ELA as an extension to Ephemeral Diffie-Hellman Over COSE (EDHOC), a key exchange protocol with extremely low message footprint. We evaluate ELA on DotBot, a platform for research in swarm micro-robotics. We find that enrolling a DotBot with ELA takes 2.52 s and consumes 39.31 mC. When compared to a baseline EDHOC version, flash and RAM have an overhead of 10.67% and 22.63%, respectively, and message footprint increases by only 49 B. ELA is being standardized in the Internet Engineering Task Force (IETF)
Stoichiometric Insights into SARS-CoV-2 Spike–ACE2 Binding Across Variants
No full textAbstract The SARS-CoV-2 spike protein binds to the angiotensin-converting enzyme 2 (ACE2) receptor to mediate viral entry, with mutations in different variants influencing binding affinity and conformational dynamics. Using large-scale molecular dynamics simulations, we analyzed the Spike–ACE2 complex in the wild-type (WT), Beta, and Delta variants. Our findings reveal significant conformational rearrangements at the inter-face in Beta and Delta compared to WT, leading to distinct interaction networks and changes in complex stability. Binding free energy analysis further highlights variant-specific differences in ACE2 affinity, with alternative binding modes emerging over the simulation. The results enhance our understanding of spike–ACE2 stoichiometry across variants, providing implications for viral infectivity and therapeutic targeting
PIA : Enseigner la protection des données personnelles dans l'interdisciplinarité
No full textNational audienceLe module d'enseignement PIA (Privacy Impact Assesment), conjuguant informatique et droit, entend créer un dialogue concret entre étudiants en informatique et étudiants en droit afin d'envisager de manière novatrice la protection des données personnelles. Associant l'INSA de Lyon, la Faculté de droit de Nantes Université ainsi que la CNIL, le module PIA, dont la première séance a été organisée en février 2025, doit permettre aux étudiants d'acquérir des connaissances techniques et juridiques centrales pour leur cursus tout en les familiarisant avec l'interdisciplinarité et la nécessité d'engager un dialogue constructif au-delà de leur domaine de compétence
Computing the dynamic response of periodic waveguides with nonlinear boundaries using the Wave Finite Element Method
No full textInternational audienceA new method to compute the dynamic response of periodic waveguides with localised nonlinearities is introduced and used to investigate the nonlinear shift of a band-edge mode in the bandgap of a locally resonant phononic structure. This nonlinear extension of the Wave Finite Element Method (WFEM) uses a finite-element discretisation of arbitrarily complex unit-cells, and leverages Floquet-Bloch theory to reduce the analysis of the entire waveguide to a state-vector of Bloch waves' amplitude. Higher harmonics generated by nonlinear effects are addressed using the Harmonic Balance Method and the nonlinear forces are evaluated via an alternating frequency-time procedure. The periodic response of the system is computed through a continuation scheme, taking the Bloch waves' amplitude as unknowns. The accuracy of the nonlinear WFEM is validated against standard FEM with Craig-Bampton reduction, demonstrating an 83% speedup in resolution time. Applying the method to a locally resonant metamaterial demonstrates that nonlinear effects can shift resonances from outside to inside bandgaps, resulting in high-amplitude, spatially localised vibrations where small amplitudes are expected from linear theory. The versatility and computational efficiency of this nonlinear dynamic simulation method should facilitate the study of complex metamaterials and civil engineering structures coupled with nonlinear interfaces or singularities