Publications

2025

• Ask and Remember: A Questions-Only Replay Strategy for Continual Visual Question Answering
  
  • Marouf Imad Eddine
  • Tartaglione Enzo
  • Lathuilière Stéphane
  • van de Weijer Joost
  , 2025, pp.1-13. Continual Learning in Visual Question Answering (VQACL) requires models to learn new visual-linguistic tasks (plasticity) while retaining knowledge from previous tasks (stability). The multimodal nature of VQACL presents unique challenges, requiring models to balance stability across visual and textual domains while maintaining plasticity to adapt to novel objects and reasoning tasks. Existing methods, predominantly designed for unimodal tasks, often struggle to balance these demands effectively. In this work, we introduce QUestion-only replay with Attention Distillation (QUAD), a novel approach for VQACL that leverages only past task questions for regularisation, eliminating the need to store visual data and addressing both memory and privacy concerns. QUAD achieves stability by introducing a question-only replay mechanism that selectively uses questions from previous tasks to prevent overfitting to the current task's answer space, thereby mitigating the out-of-answer-set problem. Complementing this, we propose attention consistency distillation, which uniquely enforces both intra-modal and inter-modal attention consistency across tasks, preserving essential visual-linguistic associations. Extensive experiments on VQAv2 and NExT-QA demonstrate that QUAD significantly outperforms state-of-the-art methods, achieving robust performance in continual VQA.
• LaCoOT: Layer Collapse through Optimal Transport
  
  • Quétu Victor
  • Liao Zhu
  • Hezbri Nour
  • Pizzati Fabio
  • Tartaglione Enzo
  , 2026, pp.20497-20507. Although deep neural networks are well-known for their outstanding performance in tackling complex tasks, their hunger for computational resources remains a significant hurdle, posing energy-consumption issues and restricting their deployment on resource-constrained devices, preventing their widespread adoption. In this paper, we present an optimal transport-based method to reduce the depth of over-parametrized deep neural networks, alleviating their computational burden. More specifically, we propose a new regularization strategy based on the Max-Sliced Wasserstein distance to minimize the distance between the intermediate feature distributions in the neural network. We show that minimizing this distance enables the complete removal of intermediate layers in the network, achieving better performance/depth trade-off compared to existing techniques. We assess the effectiveness of our method on traditional image classification setups and extend it to generative image models. Our code is available at https://github.com/VGCQ/LaCoOT. (10.1109/ICCV51701.2025.01906)
  DOI : 10.1109/ICCV51701.2025.01906
• SUB: Benchmarking CBM Generalization via Synthetic Attribute Substitutions
  
  • Bader Jessica
  • Girrbach Leander
  • Alaniz Stephan
  • Akata Zeynep
  , 2025, pp.23188-23198. Concept Bottleneck Models (CBMs) and other conceptbased interpretable models show great promise for making AI applications more transparent, which is essential in fields like medicine. Despite their success, we demonstrate that CBMs struggle to reliably identify the correct concepts under distribution shifts. To assess the robustness of CBMs to concept variations, we introduce SUB: a fine-grained image and concept benchmark containing 38,400 synthetic images based on the CUB dataset. To create SUB, we select a CUB subset of 33 bird classes and 45 concepts to generate images which substitute a specific concept, such as wing color or belly pattern. We introduce a novel Tied Diffusion Guidance (TDG) method to precisely control generated images, where noise sharing for two parallel denoising processes ensures that both the correct bird class and the correct attribute are generated. This novel benchmark enables rigorous evaluation of CBMs and similar interpretable models, contributing to the development of more robust methods. Our code is available at https://github.com/ExplainableML/sub and the dataset at http://huggingface.co/datasets/Jessica-bader/SUB. (10.1109/ICCV51701.2025.02153)
  DOI : 10.1109/ICCV51701.2025.02153
• Comment l'éthique de l'IA parle d'équité : étude linguistique de l'usage du terme "fairness" dans un corpus de chartes anglophone
  
  • Allègre Nicolas
  • Bernard Josephine
  • Hayoun Dan
  • Magne Laury
  • Viard Tiphaine
  • Boritchev Maria
  , 2025. La base de données du Conseil de l’Europe liste des thèmes et mots clé communs à travers les initiatives portant sur l’éthique de l’intelligence artificielle1. Parmi ces mots clé, nous nous sommes particulièrement intéressées aux usages du terme “fairness”. Dans la liste de thèmes citée ci-dessus, ce terme a une place particulière, notamment dans un cadre de recherche francophone; l’usage (qui fait la langue) est de parler de « la fairness » plutôt que d’utiliser une des traductions possibles (« impartialité », « équité ») lorsque l’on parle d’intelligence artificielle en français. Nous nous sommes donc attachées à explorer l’usage de “fairness” dans un contexte anglophone dans un premier temps, afin de pouvoir comparer ensuite cet usage à celui rencontré dans des contextes francophones spécialisés. Le corpus MapAIE2 est constitué de 436 chartes et manifestes de l’éthique de l’intelligence artificielle (IA), en anglais, provenant de communautés industrielles, d’entités legislatives ou politiques (ex : Commission Européenne), mais aussi de communautés scientifiques et médiatiques (Gornet et al., 2024). La sélection des documents compris dans MapAIE a été faite notamment par rapport à leur mention dans des études sociologiques précédentes; dans l’étude présentée ici, nous nous sommes appuyées sur une version non-restreinte du corpus MapAIE, ci-après dénommée SourceAIE. Cette version contient 553 documents. Nous présentons ici le résultat de notre exploration de l’usage du terme “fairness” dans le corpus SourceAIE : comment le terme “fairness” est-il utilisé dans les textes portant sur l’éthique de l’intelligence artificielle? Notre contribution est double : — Notre étude montre que le terme “fairness” apparaît principalement comme élément accessoire des phrases du corpus SourceAIE. Danslecadredenotreétude,nousavonsutiliséuneméthodeànotreconnaissance peuoupasutiliséeenlinguistiquecomputationnelle:l’étudedelaformedesgraphes AMR(AbstractMeaningRepresentation,(Banarescuetal.,2013)).Nousprésentons icinotreméthodologie.
• Bayesian Methods for Blind Beamforming in MIMO systems using Reflective Intelligent Surfaces
  
  • Chêne Thomas
  , 2025. To address the ever increasing data rate in fifth-generation (5G), multiple solutions are envisioned.Densifying the network by adding Base Stations(BS) that each have a smaller coverage allows to reuse the spectrum over a geographic area. To reduce the cost of conventional BS with high energy consumption and hardware cost, it is envisioned in Centralized Radio Access Network (CRAN) architecture to share and offload the computational power and ressources at a Central Processor(CP). The link connecting the Remote Radio Header (RRH) to the CP (referred as the fronthaul link) is likely to be overwhelmed by the data traffic since it has a limited capacity. This limits the performances of envisioned CRAN systems, and as the computational power is offloaded to the CP, RRH have to implement a low complexity compression protocol before sending their received data to the CP.Since Mmmwaves are more susceptible to blockage and absorption, the wireless environment needs to be modified in a cost-effective way.Reflective Intelligent Surfaces(RIS) are a recent technology composed of many passive reflecting elements. They are envisaged to be deployed on walls, ceilings of buildings to create a link between the User Equipment(UE) and the BS. The main challenge of RIS technology is to determine the optimal configuration of the reflecting elements. Channel acquisition for those systems poses formidable challenges as the size of the channel matrix increases with the number of passive elements, and as the RIS is passive and cannot estimate the channel.The first part of this thesis, aims at reducing the effects of a limited fronthaul capacity in a CRAN system. We model the limited capacities of the links between the RRH and the CP as a bit budget allocated to each RRH. We jointly optimize a compression protocol at the RRH and a decoder at the CP by training a neural network with a bit budget constraint.The second part of this thesis, aims at configuring a RIS to maximize the achievable rate between a BS and UE, without knowledge of the channel. We model the channel as a random variable and use an adaptive protocol that receives feedback and updates its knowledge of the channel. This bayesian method can be "accelerated" in order to reduce the number of pilots required to be sent.We first propose to efficiently "query" the channel by maximizing the amount of information each pilots brings us on relevant parameters of the channel.Then we formulate the problem as a minimization of a path length between an initial state without knowledge of the best configuration of the RIS, and a final state with almost certainty to find the best configuration.Finally, we propose some modifications to Bayesian Optimization to maximize the Received Signal Strength at the BS. (10.70675/07912ed0z295az4f70z9fabzac6728ba6ea4)
  DOI : 10.70675/07912ed0z295az4f70z9fabzac6728ba6ea4
• Mixed Criticality Mission Planning for Autonomous Robot Fleets
  
  • Cordeiro Franco Petrone
  , 2025. This thesis explores the problem of managing uncertainty in multi-robot critical systems planning. The first contribution consists of adapting Mixed-Criticality concepts from safety-critical systems to the robot planning domain. Drawing from previous work in real-time scheduling problems, this thesis reconceptualizes how robots prioritize critical tasks when resources become constrained. The approach classifies robot actions according to their objective's importance and implements multiple cost modes to handle varying environmental conditions. The second contribution is the development of a single-robot framework based on Monte-Carlo Tree-Search that demonstrates increased objective achievement in normal environments while guaranteeing critical objective execution during exceptional conditions. The third contribution is extending this solution to multi-robot systems through an approach that includes robot partitioning strategies and a robust synchronization process for online replanning, enabling robots to adapt to changing conditions in real-time. This multi-robot implementation called RESCUE tackles the additional challenge of preventing objective duplication across robots while maintaining system flexibility. This approach is also generalized to multiple levels of criticality. Finally, the contributions are evaluated through simulation by comparing them to existing Monte-Carlo Tree-Search solutions. The experimental results validate that the framework successfully balances competing priorities: maximizing objective completion during normal operation while ensuring critical task execution during environmental challenges. These contributions advance the field of adaptive planning for uncertain robotic environments with objective criticality by providing a more robust and resilient approach to resource allocation in the face of unpredictable conditions. (10.70675/ffb05b28z4782z4102zb8e6zaf93ba5fdd0b)
  DOI : 10.70675/ffb05b28z4782z4102zb8e6zaf93ba5fdd0b
• SMACC: Sketching Motion for Articulated Characters with Comics-based annotations
  
  • Legrand Amandine
  • Parakkat Amal Dev
  • Rohmer Damien
  , 2025, pp.1-13. <div><p>We introduce SMACC, a sketch-based system for animating short sequences of 3D articulated characters inspired by 2D comic motion line annotations. SMACC relies on classical rules of motion depiction used in comic books, allowing the depiction of dynamism in static images while being universally understood. Building on this, SMACC introduces an algorithmic interpretation of these principles in the context of a 3D character animation, guided by three fundamental types of motion lines: trajectory, circumfixing and impact. The adaptation to rigged 3D characters relies on the automatic computation of how these motion cues spatially influence the character's skeleton, achieved through a global analysis of sketch annotations relative to the character's pose. The resulting animation is generated by encoding the kinematic clues and constraints into joint angular velocities. Finally, the proof-of-concept demonstrated by SMACC is validated through a user study, which evaluates the effectiveness and accuracy of this sketch-based approach applied to 3D character animation.</p></div> (10.2312/pg.20251255)
  DOI : 10.2312/pg.20251255
• Asynchronous Gossip Algorithms for Rank-Based Statistical Methods
  
  • van Elst Anna
  • Colin Igor
  • Clémençon Stephan
  , 2025, pp.448-455. <div><p>As decentralized AI and edge intelligence become increasingly prevalent, ensuring robustness and trustworthiness in such distributed settings has become a critical issue-especially in the presence of corrupted or adversarial data. Traditional decentralized algorithms are vulnerable to data contamination as they typically rely on simple statistics (e.g., means or sum), motivating the need for more robust statistics. In line with recent work on decentralized estimation of trimmed means and ranks, we develop gossip algorithms for computing a broad class of rank-based statistics, including L-statistics and rank statisticsboth known for their robustness to outliers. We apply our method to perform robust distributed two-sample hypothesis testing, introducing the first gossip algorithm for Wilcoxon rank-sum tests. We provide rigorous convergence guarantees, including the first convergence rate bound for asynchronous gossip-based rank estimation. We empirically validate our theoretical results through experiments on diverse network topologies.</p></div> (10.1109/FLTA67013.2025.11336445)
  DOI : 10.1109/FLTA67013.2025.11336445
• Secure implementation for post-quantum cryptography
  
  • Spyropoulos Maxime
  , 2025. The goal of the thesis is to improve the software security of components implementing post-quantum cryptography. More specifically, the aim is to identify and correct vulnerabilities to auxiliary channel attacks. (10.70675/2a5c3019zb14cz41bdzbe8bz42175951619c)
  DOI : 10.70675/2a5c3019zb14cz41bdzbe8bz42175951619c
• Physically Informed Spatial Regularization for Sound Event Localization and Detection
  
  • Liu Haocheng
  • Di Carlo Diego
  • Nugraha Aditya Arie
  • Yoshii Kazuyoshi
  • Richard Gaël
  • Fontaine Mathieu
  , 2025. Building Sound Event Localization and Detection (SELD) models that are robust to diverse acoustic environments remains one of the major challenges in multichannel signal processing, as reflections and reverberation can significantly confuse both the source direction and event detection. Introducing priors such as microphone geometry or room impulse response (RIR) into the model has proven effective in addressing this issue. Existing methods typically incorporate such priors in a deterministic way, often through data augmentation to enlarge data diversity. However, the uncertainty arising from the complex nature of audio acoustics remains largely underexplored in the SELD literature and naturally call for incorporating a stochastic modeling of acoustic prior. In this paper, we propose regularizing deep learning based SELD models with a physically constructed spatial covariance matrix (SCM) based on the estimated direction of arrival (DOA) and sound event detection (SED).
• IS³ : Generic Impulsive--Stationary Sound Separation in Acoustic Scenes using Deep Filtering
  
  • Berger Clémentine
  • Stamatiadis Paraskevas
  • Badeau Roland
  • Essid Slim
  , 2025. We are interested in audio systems capable of performing a differentiated processing of stationary backgrounds and isolated acoustic events within an acoustic scene, whether for applying specific processing methods to each part or for focusing solely on one while ignoring the other. Such systems have applications in real-world scenarios, including robust adaptive audio rendering systems (e.g., EQ or compression), plosive attenuation in voice mixing, noise suppression or reduction, robust acoustic event classification or even bioacoustics. To this end, we introduce IS³, a neural network designed for Impulsive--Stationary Sound Separation, that isolates impulsive acoustic events from the stationary background using a deep filtering approach, that can act as a pre-processing stage for the above-mentioned tasks. To ensure optimal training, we propose a sophisticated data generation pipeline that curates and adapts existing datasets for this task. We demonstrate that a learning-based approach, build on a relatively lightweight neural architecture and trained with well-designed and varied data, is successful in this previously unaddressed task, outperforming the Harmonic--Percussive Sound Separation masking method, adapted from music signal processing research, and wavelet filtering on objective separation metrics.
• Meaning Representation Frameworks and Reasoning in the Era of Large Language Models
  
  • Sadeddine Zacchary
  , 2025. Large Language Models (LLMs) are now used for a wide range of tasks, many of which require reasoning abilities. However, these abilities remain limited and lack transparency. This thesis explores how to improve the reasoning, transparency and robustness of LLMs by integrating symbolic structures.First, we conduct an analysis of the societal issues arising from the new role of LLMs in our access to knowledge. Fifteen major issues are identified, as well as current and potential mitigation strategies, drawing on both technical solutions and regulatory approaches.The thesis then focuses on Meaning Representation Frameworks (MRFs), which encode the semantics of natural language into graph structures. A comprehensive survey of MRFs is presented, introducing a new classification based on their structural properties, as well as the available resources, empirical use and research directions. This repositions MRFs as computational artifacts capable of complementing neural models in complex tasks.Building upon this, the thesis introduces VANESSA, a neuro-symbolic reasoning system that integrates MRFs with LLMs, as well as new representation and symbolic parsing process. VANESSA uses this representation to decompose reasoning problems into three simpler subtasks: parsing, natural language inference (NLI) and formal solving. Experimental results show that VANESSA achieves performance comparable to LLMs on logical reasoning tasks, while producing outputs that are traceable and explainable, illustrating the added value of hybrid architectures.Finally, the thesis addresses the problem of step-by-step verification of reasoning chains, which are produced by LLMs. A novel benchmark of nearly 5,000 annotated reasoning steps is presented, assessing both logical validity and factual correctness. Though LLMs are able to detect some errors, neuro-symbolic approaches such as VANESSA achieve comparable performance while providing valuable transparency.Overall, the thesis advocates a hybrid vision of language-based artificial intelligence, where LLMs and symbolic structures are not competing paradigms but complementary tools. It opens new perspectives towards AI systems that are not only powerful, but also responsible, trustworthy and interpretable, combining the flexibility of neural models with the rigor of symbolic reasoning. (10.70675/7f1c15a9z3c9bz4a61z8bf8z8bc5790ee4ed)
  DOI : 10.70675/7f1c15a9z3c9bz4a61z8bf8z8bc5790ee4ed
• Photonic Chaos in Quantum Cascade Lasers : Foundations and Applications in Free-Space Optical Systems
  
  • Zaminga Sara
  , 2025. This doctoral thesis explores the use of chaotic light for next-generation free-space optical (FSO) communication systems, focusing on quantum cascade lasers (QCLs) operating in the long-wave infrared (LWIR) atmospheric window. At the core of the study are distributed-feedback (DFB) QCLs, whose unique dynamics are investigated using the Effective Semiconductor Maxwell-Bloch Equations (ESMBEs).We reveal how physical effects—such as a non-zero linewidth enhancement factor (LEF) and fast spatial hole burning (SHB)—alongside geometrical factors like cavity length and facet coatings, govern both the spectral stability and intrinsic modulation response. These mechanisms are critical to understanding the transition from single-mode to multimode longitudinal emission as the bias current increases.In the presence of external optical feedback, we show that photonic chaos emerges through the interplay between internal longitudinal modes and external cavity modes—not from undamped relaxation oscillations, as in interband lasers. The onset of chaos requires feedback strengths nearly two orders of magnitude higher than in diode lasers, consistent with the quasi-Class A nature of QCLs.Building on this insight, we demonstrate two pioneering applications. First, we realize the first LWIR chaos-based LiDAR system, achieving sub-centimeter precision and meter-range resolution—currently limited by detector bandwidth. Second, we present a chaos-based random number generator (RNG) using DFB QCLs, reaching bit-rates up to 2.5 Gbps—marking a first in this spectral region.We further examine the resilience of chaotic signals against atmospheric turbulence in the C-band, at 1.55 µm. Using a spatial light modulator to emulate turbulence in the laboratory environment and a self-configurable programmable photonic processor at the receiver end, we recover the degraded chaotic dynamics due to propagation through a turbulent medium, validating the feasibility of turbulence-hardened FSO links.This work lays the foundation for a new class of LWIR photonic systems that harness deterministic chaos as a resource. By bridging advanced laser physics, nonlinear dynamics, and real-world applications, it paves the way for high-speed, secure, and turbulence-resilient FSO technologies—unlocking new possibilities in remote sensing, telecommunications, and information security. (10.70675/1aa6d967z4564z46b6z94c3ze7d1c1e04e1a)
  DOI : 10.70675/1aa6d967z4564z46b6z94c3ze7d1c1e04e1a
• Phase Diagram of Dropout for Two-Layer Neural Networks in the Mean-Field Regime
  
  • Chizat Lénaïc
  • Marion Pierre
  • Yesbay Yerkin
  , 2025. Dropout is a standard training technique for neural networks that consists of randomly deactivating units at each step of their gradient-based training. It is known to improve performance in many settings, including in the large-scale training of language or vision models. As a first step towards understanding the role of dropout in large neural networks, we study the large-width asymptotics of gradient descent with dropout on two-layer neural networks with the mean-field initialization scale. We obtain a rich asymptotic phase diagram that exhibits five distinct nondegenerate phases depending on the relative magnitudes of the dropout rate, the learning rate, and the width. Notably, we find that the well-studied "penalty" effect of dropout only persists in the limit with impractically small learning rates of order O(1/width). For larger learning rates, this effect disappears and in the limit, dropout is equivalent to a "random geometry" technique, where the gradients are thinned randomly after the forward and backward pass have been computed. In this asymptotic regime, the limit is described by a mean-field jump process where the neurons' update times follow independent Poisson or Bernoulli clocks (depending on whether the learning rate vanishes or not). For some of the phases, we obtain a description of the limit dynamics both in path-space and in distribution-space. The convergence proofs involve a mix of tools from mean-field particle systems and stochastic processes. Together, our results lay the groundwork for a renewed theoretical understanding of dropout in large-scale neural networks.
• TTool-AI: A Large Language Model-Based Assistant for Model Driven Engineering
  
  • Sultan Bastien
  • Apvrille Ludovic
  SN Computer Science, Springer, 2025, 6 (7), pp.886 (1-18). Throughout the history of engineering, successive innovations have been implemented to assist engineers in their tasks, enabling them to focus on high-value activities while minimizing time-consuming and error-prone tasks. Large language models (LLMs) represent one of these innovations, with significant potential for developing new kinds of engineering assistants, as demonstrated by a rich body of recent literature. The paper introduces TTool-AI, a model-driven engineering assistant based on LLMs and integrated within the SysML modeling and formal verification toolkit TTool. TTool-AI enables system architects to generate and incrementally refine various types of SysML diagrams directly from textual specifications with a single click. The core mechanisms of TTool-AI (contextual knowledge injection, automated prompt generation, and iterative feedback) enable it to produce good quality models that can serve as a sound foundation for system architects in MDE processes. Building on our previous work presented at MODELSWARD 2024, this paper provides a comprehensive description of TTool-AI’s MDE assistance features. It introduces new functionalities, including requirement engineering and automated model mutation generation. An evaluation of these features, comparing their performance against Master-level students, demonstrates the tool’s efficacy and suggests a strong potential to significantly enhance engineering productivity by enabling engineers to focus on high-value tasks. (10.1007/s42979-025-04444-w)
  DOI : 10.1007/s42979-025-04444-w
• Numerically Efficient Parametric Inference for Learning Space-Time Hawkes Processes
  
  • Siviero Emilia
  • Staerman Guillaume
  • Clémençon Stéphan
  • Moreau Thomas
  , 2025, pp.1-10. In a wide range of spatio-temporal datasets, from sociology to seismology, self-exciting dynamics are often observed, characterized by event triggering and clustering across both space and time. Space-time Hawkes processes provide a powerful framework to model such phenomena. This paper introduces a flexible parametric inference method to estimate the underlying kernel parameters involved in the intensity function of a space-time Hawkes process based on such data. Our approach combines three core components: 1) kernels with finite support, 2) discretization of the space-time domain, and 3) efficient (possibly approximate) precomputations. The inference method we propose then relies on a gradient-based solver that offers both computational efficiency and strong statistical performance. Alongside a detailed presentation of the algorithmic framework, we present numerical experiments on synthetic and real spatio-temporal data, offering solid empirical evidence of the validity and applicability of the proposed methodology. (10.1109/DSAA65442.2025.11247997)
  DOI : 10.1109/DSAA65442.2025.11247997
• How to Improve Anomaly Detection for Electric Powertrains in Production?
  
  • Emelchenkov Anton
  • Fontaine Mathieu
  • Mahé Hervé
  • Roueff François
  , 2025. Despite the low noise level of an electric powertrain, its tonality concentrated around a few frequencies can make it painful for the end user. The End Of Line Tester (EOLT) for electric powertrains plays a critical role in ensuring NVH quality standards. Today’s industry-standard solutions predominantly rely on order tracking and amplitude estimation to detect potential defects and compliance versus requirements. These techniques often depend on expert intervention and precise hyperparameters tuning, which undermines their robustness and scalability, especially when faced with rapidly evolving non-stationary signals. To reinforce precision and speed, two key innovations are proposed: (1) a high-resolution method for multi-frequency amplitude estimation in highly oscillatory regimes, equipped with automatic hyperparameter tuning to enhance the accuracy and stability of order tracking; and (2) a neural network-based anomaly detection framework that learns directly from raw signal spectrograms, removing the need for handcrafted signal processing. To support this, we introduce and release the first dataset of non-stationary vibration signals collected from an EOLT, specifically designed for anomaly detection. Our approach sets a new benchmark for automated, data-driven diagnostics in electric powertrain manufacturing.
• Adaptive Augmented Reality Pathfinding for Parkinson's Disease: Integrating Visual Cueing with User-Directed Navigation
  
  • Bassal Dimah
  • Medeiros Daniel
  , 2025. <div><p>Parkinson's disease (PD) affects millions worldwide, with gait impairments and freezing of gait (FOG) episodes representing debilitating symptoms that significantly compromise patient mobility, safety, and quality of life. Recent advances in augmented reality (AR) have demonstrated that visual cueing can effectively modify gait parameters in PD patients, with AR cues proving as effective as real-world cues for improving step length, gait speed, and crossing maneuvers. However, existing AR cueing systems, including applications like Holocue, rely primarily on static, pre-positioned cues or continuous cueing paradigms that require patients to adapt to predetermined patterns, poten-tially limiting patient autonomy and confidence in independent navigation. This paper presents a novel adaptive AR pathfinding system that integrates established visual cueing principles with user-directed navigation to enhance both therapeutic effectiveness and patient autonomy.</p></div>
• Associations between individual and geospatial characteristics and power of 4G signals received by mobile phones
  
  • Laplanche Alexia
  • Guida Florence
  • Moissonnier Monika
  • Launay Ludivine
  • Beranger Remi
  • Lagroye Isabelle
  • Orlacchio Rosa
  • Fontaine Maëlle
  • Bories Serge
  • Mazloum Taghrid
  • Conil Emmanuelle
  • Huss Anke
  • Wiart Joe
  • Danjou Aurélie
  • Schüz Joachim
  • Dejardin Olivier
  • Deltour Isabelle
  Environmental Research, Elsevier, 2025, 286 (3), pp.123030-1:123030-11. Background: The Received Signal Strength Indicator (RSSI) measures downlink signal intensity received by smartphones in 4th Generation LTE networks. Objective: This study evaluated how individual, technical, and spatial factors influenced LTE-RSSI during daily activities. Methods: Between November 2022 and October 2023, adults in France used the XMobiSensePlus Android smartphone application to record RSSI and GPS data. Distance to the operator's nearest antenna, obtained from Cartoradio, population and antenna density and urbanicity were analyzed using a geographic information system. Determinants of RSSI were assessed using an autoregressive mixed model incorporating restricted cubic splines for distance. Environmental exposures were estimated at 1800 MHz using conversion factors. Results: From 1,969,913 records of 187 participants, with measurements taken every 30 s over 7.9 days, the average LTE-RSSI was -79.3 dBm. The estimated electric field strength was 0.12 V/m, albeit with large uncertainty. The median distance to the nearest antenna was 536 m. Proximity to antennas increased RSSI. Antenna density positively influenced RSSI (overall β = +0.37 dBm per additional antenna per km2). Lower RSSI was observed in the evening and night, particularly in urban areas. Smartphone's technical parameters (Android version and System-on-a-Chip) influenced RSSI, operators did not. Proximity to antennas had greater impact in rural areas. Conclusion: Urbanicity, distance to the nearest 4G antenna, antenna density, time of day, and smartphone's technical parameters influenced RSSI levels in 4G networks in France, but not operator. (10.1016/j.envres.2025.123030)
  DOI : 10.1016/j.envres.2025.123030
• Fine-Grained Confidentiality and Authenticity Modeling and Verification for Embedded Systems
  
  • Jerray Jawher
  • Sultan Bastien
  • Apvrille Ludovic
  , 2025, pp.333-344. Handling cybersecurity during system design is mandatory for (critical and) connected embedded systems. Numerous contributions, including standards like ISO 26262, emphasize the need to address cybersecurity as early as possible in the design process. Design space exploration, typically performed early in system design-before software or hardware development-offers an opportunity for early cybersecurity integration. SysML-Sec has demonstrated how cybersecurity concepts can be incorporated into design space exploration. However, its security mechanisms have significant limitations to address some of the modern threats. The paper introduces a new security modeling and verification approach. Our method enables multipattern security channels, allowing multiple security patterns to coexist within a single communication channel. It also supports fine-grained verification of individual write and read operations, ensuring that confidentiality and authenticity are independently validated for each data exchange. Additionally, our approach generates traceable counterexamples for unverified properties, helping engineers identify and address security vulnerabilities. We implemented this technique in TTool/DIPLODOCUS, a UML/SysML-based framework for hardware/software co-design, demonstrating how its enhanced version can now support more advanced security mechanisms, and evaluated it on an automotive case-study. (10.1109/MODELS-C68889.2025.00052)
  DOI : 10.1109/MODELS-C68889.2025.00052
• Tunable Metasurface MIMO Antenna
  
  • Medrar Ghiles
  • Lepage Anne Claire
  • Begaud Xavier
  , 2025.
• Quantitative Limit Theorems for Cox-Poisson and Cox-Binomial Point Processes
  
  • Adrat Hamza
  • Decreusefond Laurent
  , 2025. <div><p>This paper establishes quantitative limit theorems for two classes of Cox point processes, quantifying their convergence to a Poisson point process (PPP). We employ Stein's method for PPP approximation, leveraging the generator approach and the Stein-Dirichlet representation formula associated with the Glauber dynamics. First, we investigate a Cox-Poisson process constructed by placing one-dimensional PPPs on the lines of a Poisson line process in $\mathbb{R}^2$. We derive an explicit bound on the convergence rate to a homogeneous PPP as the line intensity grows and the point intensity on each line diminishes. Second, we analyze a Cox-Binomial process on the unit sphere $\mathbb{S}^2$, modeling a system of satellites. This process is generated by placing PPPs on great-circle orbits, whose positions are determined by a Binomial point process. For this model, we establish a convergence rate of order O(1/n) to a uniform PPP on the sphere, where n is the number of orbits. The derived bounds provide precise control over the approximation error in both models, with applications in stochastic geometry and spatial statistics.</p></div>
• SuperviZ - Supervision et orchestration de la sécurité - Rapport d’avancement à mi-projet
  
  • Debar Hervé
  • Mé Ludovic
  • Leneutre Jean
  • Nicomette Vincent
  • François Jérôme
  • Gouy-Pailler Cédric
  • Blanc Gregory
  • Mocanu Stéphane
  , 2025, pp.1-58. Ce document constitue le rapport à mi-parcours du projet SuperviZ. Il regroupe l’ensemble des livrables à mi-parcours des 6 lots du projet (un chapitre par lot), identifiés sous les codes L02 à L07. Le projet SuperviZ s’intéresse à la détection, à la réponse et à la remédiation des attaques informatiques, sujets regroupés sous l’appellation de “supervision de sécurité”. Fondamentale dans le contexte des SI d’entreprise, la supervision l’est encore plus dans le cas des systèmes cyber-physiques. En effet, avec des “objets” (dispositifs de nature et de capacité très hétérogène) qui devraient à terme être tous, ou presque, connectés, la surface d’attaque augmente significativement. Le projet SuperviZ adresse des défis de la supervision dans ce double contexte IT et OT. Il a permis de lancer à ce jour 13 thèses (pour 11 initialement prévues) et 5 postdoc (2 sont terminés et 2 restent à pourvoir).
• Efficient Quantum Measurements: Computational Max-and Measured Rényi Divergences and Applications
  
  • Yángüez Álvaro
  • Hahn Thomas A
  • Kochanowski Jan
  , 2025. Quantum information processing is limited, in practice, to efficiently implementable operations. This motivates the study of quantum divergences that preserve their operational meaning while faithfully capturing these computational constraints. Using geometric, computational, and information theoretic tools, we define two new types of computational divergences, which we term computational max-divergence and computational measured Rényi divergences. Both are constrained by a family of efficient binary measurements, and thus useful for state discrimination tasks in the computational setting. We prove that, in the infinite-order limit, the computational measured Rényi divergence coincides with the computational max-divergence, mirroring the corresponding relation in the unconstrained information-theoretic setting. For the many-copy regime, we introduce regularized versions and establish a one-sided computational Stein bound on achievable hypothesis-testing exponents under efficient measurements, giving the regularized computational measured relative entropy an operational meaning. We further define resource measures induced by our computational divergences and prove an asymptotic continuity bound for the computational measured relative entropy of resource. Focusing on entanglement, we relate our results to previously proposed computational entanglement measures and provide explicit separations from the information-theoretic setting. Together, these results provide a principled, cohesive approach towards state discrimination tasks and resource quantification under computational constraints.
• Image Pre-Segmentation from Shadow Masks
  
  • Heep Moritz
  • Parakkat Amal Dev
  • Zell Eduard
  , 2025, pp.1-7. Image segmentation has gained a lot of attention in the past. When working with photometric stereo data, we discovered that shadow cues provide valuable spatial information, especially when combining multiple images of the same scene under different lighting conditions. In the following, we present a robust method to pre-segment images, relying heavily on shadow masks as the main input. We first detect object contours from light to shadow transitions. In the second step, we run an image segmentation algorithm based on Delaunay triangulation that is capable of closing the gaps between contours. Our method requires spatial input data but is free from training data. Initial results look promising, generating pre-segmentations close to recent data-driven image segmentation algorithms. (10.2312/vmv.20251239)
  DOI : 10.2312/vmv.20251239