Archive ouverte de Centrale Lyon
Not a member yet
32420 research outputs found
Sort by
An Innovative Data Mining Technique for Automatic Anomaly Detection in Physical Unclonable Functions
No full textInternational audiencePhysical Unclonable Functions (PUFs) offer a promising alternative to conventional cryptographic techniques to secure sensitive information in modern circuits. PUFs leverage inherent process variability to dynamically generate unique secrets, eliminating the need for data storage. However, a significant challenge in PUF-based security is differentiating between valid PUFs and those that may have been tampered with or are invalid (i.e., not belonging to the original design). This paper presents an innovative data mining-based technique for detecting anomalies and identifying tampered or invalid PUFs. The proposed method extracts a set of rules that describe the expected behavior of the PUF, where deviations from these rules indicate potential security issues and vulnerabilities. Experimental results demonstrate that the method effectively detects invalid or tampered PUFs, highlighting its potential to strengthen PUF-based security systems
Démarches du prototypage rapide de la commande d'un onduleur à commutation douce
No full textInternational audienceLes onduleurs à commutations douces sont de plus en plus investigués afin d'augmenter la densité de puissance globale de la chaine de conversion électromécanique. L'utilisation de la modulation de largeur d'impulsion (MLI) est aujourd'hui une technique indispensable pour générer les commandes des convertisseurs électroniques de puissance. Elle permet d'assurer un contrôle précis des applications nécessitant une variation de fréquence avec une densité de puissance élevée. Dans cet article, nous abordons le développement des lois de commande de la MLI pour un onduleur monophasé à commutation douce, et leurs implémentations sur une cible FPGA (Field Programmable Gate Array), permettant un contrôle rapide en temps réel
generic-query-modify
No full textProgram for the paper "A Generic Query-Modify Framework for Volumetric Mesh Processing". Guillaume Damiand, Vincent Nivolliers, Romain Pascual.This repository contains code, scripts and dataset.The paper introduces a query-modify framework for automating volumetric mesh processing. Our method enables flexible and efficient modifications of geometric structures composed of multiple volumes with minimal user-implemented code. Modifications are provided as rules consisting of a query mesh and a target mesh representing structural information to be extracted and replaced. The rules enable both localized queries to be matched with a portion of an input mesh and targeted modifications on the matched portion of the input mesh. Our approach generalizes standard mesh manipulations and adapts to various applications, including geometric modeling, remeshing, and topology-aware transformations
From Human Attention to Computational Attention: A Multidisciplinary Approach
No full textInternational audienceThe new edition of this popular book introduces the study of attention, focusing on attention modeling, and addressing such themes as saliency models, signal detection, and different types of signals, including real-life applications. The first edition was written at a moment when the Deep Learning Neural Network (DNNs) techniques were just at their beginnings in terms of attention. Deep learning has recently become a key factor in attention prediction on images and video, and attention mechanisms have become key factors in deep learning models. The second edition tackles the arrival of DNNs for attention computing in images and video, and also discusses the attention mechanisms within DNNs (attention modules, transformers, grad-cam-based saliency maps, etc.). From Human Attention to Computational Attention 2nd Edition also explores the parallels between the brain structures and the DNN architectures to reveal how biomimetics can improve the model designs. The book is truly multi-disciplinary, collating work from psychology, neuroscience, engineering, and computer science
Prise en compte du vieillissement des batteries lithium-ion dans un modèle d’analyse du cycle de vie environnementale
No full textInternational audienceCe travail explore l’influence de l'intégration du vieillissement des batteries sur leurs impacts environnementaux à travers l’utilisation de la méthode d’analyse du cycle de vie (ACV) environnementale au moyen d’un outil dédié et accessible par la communauté. Dans un premier temps, une présentation de la méthodologie de l’ACV environnementale et un état de l’art contextualisant les enjeux liés à cette étude sont exposés. Ensuite, la méthodologie de modélisation et de l’estimation du vieillissement des batteries Li-ion est détaillée. La démarche pour réaliser une ACV environnementale de batterie est décrite et une quantification de l’impact du vieillissement sur les résultats d’ACV environnementale est effectuée
Weak solutions for a singular beam equation: Weak solutions for a singular beam equation
No full textInternational audienceThis paper deals with a dynamic Gao beam of innite length subjected to a moving concentrated Dirac mass. Under appropriate regularity assumptions on the initial data, the problem possesses a weak solution which is obtained as the limit of a sequence of solutions of regularized problems
Team INSALyon2 at SemEval-2025 Task 10: A Zero-shot Agentic Approach to Text Classification
No full textInternational audienceWe present Team INSALyon2's agentic approach to SemEval-2025 Task 10 Subtask 2, focusing on multi-label classification of narratives in news articles. Our system employs specialized Large Language Model agents for binary classification of individual narrative labels, with a meta-agent aggregating these decisions into final multi-label predictions. Using Auto-Gen to orchestrate GPT-based agents without fine-tuning, our approach effectively handles the two-level taxonomy classification challenge. Experiments on the English subset demonstrate competitive performance (F1 macro coarse = 0.513, F1 sample = 0.406), securing third place in the competition and showing the effectiveness of zero-shot agentic approaches for complex classification tasks
Low-Frequency Magnetic Incremental Permeability Under Rotational Magnetization for Mechanical Stress Characterization
No full textInternational audienceThis study investigates the Low-Frequency Magnetic Incremental Permeability (LF-MIP) non-destructive testing technique under rotational magnetization for evaluating internal mechanical stress. Operating in the highamplitude range minimizes the influence of ferromagnetic domain behavior, focusing the magnetization process predominantly on rotational mechanisms. These optimal conditions facilitate a more direct and reliable assessment of internal stress
Advanced Malware Detection in Code Repositories Using Graph Neural Networks
No full textInternational audienceThe proliferation of online code repositories, such as GitHub and GitLab, has increased security risks related to the spread of malware through source code, including compromised libraries and obfuscated code. Traditional detection methods, such as static and dynamic analysis, have shown limitations when faced with advanced threats. To address this issue, we propose an approach based on Graph Neural Networks (GNN) to enhance malware detection in code repositories. Graphs provide an effective way to model the complex relationships within a program, such as interactions between functions, classes, modules, and external dependencies. This is particularly useful for identifying specific patterns of complex malicious behavior. Our model detects anomalies by analyzing the program internal interactions in addition to the exact code content. Our method extracts call graphs and metadata (such as descriptions and README files) from code repositories. The metadata is converted into vectors using a natural language embedding model, then merged with the embeddings of call graphs generated by GNNs. This combined representation is used by a supervised classifier to detect malicious repositories. The results show strong performance in precision, recall, and F1-score, validating the effectiveness of our approach