Publications
-
Towards Context-Driven Approaches to Visual Affordance Recognition : A Comprehensive Survey
With the rise of robotics, affordance detection has become a key element for improving and strengthening human and robot object interactions. Thanks to recent progress in deep learning, affordance detection has become more precise and versatile. However, with the increasing use of Large Language Models (LLMs) and Visual Transformers (ViTs), the research focus shifted from […]
-
Consensus de partitions pour une sélection de variables non biaisée par forêts aléatoires
En apprentissage supervisé, la forêt aléatoire est un modèle de référence qui s’accommode de la plupart des types de données et capture naturellement interactions et non-linéarités. Reposant sur l’agrégation de nombreux arbres, elle se comporte toutefois comme une boîte noire. Les mesures d’importance des variables (VIM) restaurent l’explicabilité, mais deviennent biaisées en présence de multicolinéarité, […]
-
FLASH: Efficient Impact Fall Detection with Unified Hypergraph State-Space Model
Falls represent a critical public health challenge, and accurate detection of the impact moment when an individual hits the ground is crucial for timely intervention. Existing skeleton-based methods rely on graph neural networks modeling only pairwise joint connections, failing to capture multi-joint coordination characteristic of fall impacts, while transformer-based temporal models suffer from quadratic complexity […]
-
Designed interfaces in adhesively bonded aluminium sandwich structures with 3D-printed polymer cores
This study investigates the development of lightweight aluminium–polymer sandwich structures for aeronautical drilling applications, with particular emphasis on the mechanical performance of adhesively bonded interfaces. The sandwich structures consist of aluminium skins bonded to a 3D-printed polyethylene terephthalate glycol (PETG) core using a two-part structural epoxy adhesive. The polymer core architecture was first optimized using […]
-
Momentum-Guided Federated Split Distillation for Personalized Temporal Edge Intelligence
We propose a momentum-guided federated split distillation framework for personalized, efficient, and autonomous temporal edge intelligence. We introduce TeRR-SAtt, our novel temporal reservoir student attention design that combines fixed reservoir representations, a lightweight temporal student, and personalized output modules. We also present AMGF, our anticipatory momentum-guided fusion mechanism that clusters clients through learning momentum and […]
-
Computational Efficiency through Momentum-Guided Split Distillation: Rethinking Federated Learning for Edge Autonomy
IoT ecosystems increasingly require adaptive intelligence under heterogeneous data and resource conditions [1]. In such distributed, privacy-sensitive, and evolving environments, Federated Learning (FL) enables collaborative edge intelligence without transmitting raw data [2]. However, IoT intelligence requires temporal reasoning to capture evolving and context-dependent dynamics [3]. This makes temporal reasoning valuable yet costly for constrained devices. […]
-
Graph Neural Network Approach for Boundary Node Detection in Wireless Sensor and IoT Networks Using the LPCN Algorithm
Boundary node detection in wireless sensor networks (WSNs) and Internet of Things (IoT) networks is a fundamental problem with applications in perimeter surveillance, coverage verification, and network management. The Least Polar-angle Connected Node (LPCN) algorithm provides an exact solution by tracing the polygon hull of a connected Euclidean graph, but its distributed version (D-LPCN) requires […]
-
Forecasting short-term availability of bikes at sharing stations : a deep-learning approach
The growing popularity of bike-sharing systems represents a significant urban mobility trend, offering substantial economic, environmental, territorial and social benefits. The achievement of these advantages is contingent on the operational efficiency of the service. In this paper, the short-term bike availability forecasting at the station level is addressed using real-time open data from public bike […]
-
Multi-criteria Evaluation of Digital Twins for Industry 5.0: Sustainability, Resilience and Human-Centricity
Digital twins (DTs) are a key enabler of Industry 5.0’s objective to reconcile operational performance with sustainability and human well-being. However, there is no widely adopted and reproducible evaluation framework for assessing the contributions of DT to these objectives. To address this gap, we first conducted a systematic literature review to identify current practices and […]
-
AI-Based Energy Management of Grid Connected PV Systems in Positive Energy Buildings
In response to the growing importance of renewable energy in the global energy transition, this study addresses the critical need for efficient, real-time energy management in positive energy buildings by focusing on ultra-short-term prediction with a 5-minute prediction horizon. Utilizing a novel application of AI-driven dynamic control (AIDC) for intelligent real-time control, we employ the […]
-
Hybrid flax/carbon bonded composite patches for strengthening of steel plates: layup and adhesive effect
Adhesive bonding of Fiber-Reinforced Polymer (FRP) patches is increasingly used to strengthen steel structures. Considering that carbon FRP (CFRP) and epoxy adhesives are the primary materials in industrial applications, this study explores the feasibility of hybridizing CFRPs with Flax FRPs (FFRPs) and the effects of using different adhesives on the mechanical performance of reinforced steel […]
-
Mind the Domain Gap: Benchmarking Structure-from-Motion Methods on Paired Industrial Physical-Digital Twin Datasets to Enable Radiance Field Rendering
As industry transitions toward fully digitalized factories, mirroring physical environments in digital-twins has become essential for robotics, monitoring, and process automation, a capability that fundamentally relies on 6D camera pose estimation and 3D scene reconstruction. We propose an evaluation methodology to benchmark these tasks in paired physical and digital environments. Real-world acquisitions are performed on […]