Matteo Sammarco

Matteo Sammarco

Auto-generated profile, based on publicly available information

Matteo Sammarco

AXA Group, Paris, France

Research Focus & Metrics

Academic Career Length

8Years (2019-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
13 June 2022 in IEEE Transactions on Intelligent Vehicles
Verified
Journal Article
Closed access
FOFernando Molano Ortiz
Fernando Molano Ortiz
MSMatteo Sammarco
Matteo Sammarco
LCLuís Henrique Maciel Kosmalski Costa
Luís Henrique Maciel Kosmalski Costa
MDMarcin Detyniecki
Marcin Detyniecki
Modern vehicles are equipped with a myriad of sensors. Proprioceptive sensors monitor the vehicle status and operation, whereas exteroceptive ones sense the external environment, i.e., outside the vehicle’s mechanics domain. A myriad of vehicular applications, from monitoring to autonomous driving, rely on the information provided by sensors. In this paper, we focus those which depend on exteroceptive sensors, which are critical for modern Advanced Driving Assistance Systems (ADAS) and autonomous vehicles, regardless of their automation level. Initially, we provide an overview of vehicular sensors and Off-the-Shelf (OTS) devices. Then, we focus on exteroceptive sensors, identifying different technologies, usages, pros and cons, as well as their availability in existing OTS devices. Next, we review the services and applications using exteroceptive sensors, especially regarding safety, navigation, and driving profiling. Besides referencing the most relevant literature on exteroceptive sensors, we also identify the datasets available in the literature. We conclude with open challenges using exteroceptive sensors from both industry and research perspectives.

3 July 2023 in Accident Analysis & Prevention
Verified
Journal Article
Closed access
FOFernando Molano Ortiz
Fernando Molano Ortiz
MSMatteo Sammarco
Matteo Sammarco
MDMarcin Detyniecki
Marcin Detyniecki
LCLuís Henrique Maciel Kosmalski Costa
Luís Henrique Maciel Kosmalski Costa

24 July 2024 in Engineering Applications of Artificial Intelligence
Verified
Journal Article
Closed access
LGLuana Gantert
Luana Gantert
TZTrevor Zeffiro
Trevor Zeffiro
MSMatteo Sammarco
Matteo Sammarco
MCMiguel Elias Campista
Miguel Elias Campista

17 December 2019
Verified
Book Chapter
Closed access
PDF file
BRBoris Ruf
Boris Ruf
MSMatteo Sammarco
Matteo Sammarco
JAJonathan Aigrain
Jonathan Aigrain
MDMarcin Detyniecki
Marcin Detyniecki

4 March 2024 in Journal of Internet Services and Applications
Verified
Journal Article
Open access
PDF file
LSLucas Airam C. De Souza
Lucas Airam C. De Souza
GCGustavo F. Camilo
Gustavo F. Camilo
GRGabriel Rebello
Gabriel Rebello
MSMatteo Sammarco
Matteo Sammarco
MCMiguel Elias Campista
Miguel Elias Campista
LCLuís Henrique Maciel Kosmalski Costa
Luís Henrique Maciel Kosmalski Costa
Federated learning (FL) is a distributed approach to train machine learning models without disclosing private data from participating clients to a central server. Nevertheless, FL training struggles to converge when clients have distinct data distributions, which leads to an increased training time and model prediction error. We propose ATHENA-FL, a federated learning system that considers clients with heterogeneous data distributions to generate accurate models in fewer training epochs than state-of-the-art approaches. ATHENA-FL reduces communication costs, providing an additional positive aspect for resource-constrained scenarios. ATHENA-FL mitigates data heterogeneity by introducing a preliminary step before training that clusters clients with similar data distribution. To handle that, we use the weights of a locally trained neural network used as a probe. The proposed system also uses the one-versus-all model to train one binary detector for each class in the cluster. Thus, clients can compose complex models combining multiple detectors. These detectors are shared with all participants through the system's database. We evaluate the clustering procedure using different layers from the neural network and verify that the last layer is sufficient to cluster the clients efficiently. The experiments show that using the last layer as input for the clustering algorithm transmits 99.68% fewer bytes to generate clusters compared to using all the neural network weights. Finally, our results show that ATHENA-FL correctly identifies samples, achieving up to 10.9% higher accuracy than traditional training. Furthermore, ATHENA-FL achieves lower training communication costs compared with MobileNet architecture, reducing the number of transmitted bytes between 25% and 97% across evaluated scenarios.

31 March 2026
Verified
Thesis
Closed access
MSMatteo Sammarco
Matteo Sammarco
Dissémination multi-contenus opportuniste : monitorage passif et adaptation aux conditions du réseau: La pénétration du marché des appareils mobiles a connu une croissance impressionnante ces dernières années. Smartphones, tablettes et ordinateurs portables sont devenus soit producteurs soit consommateurs de contenus générés par les utilisateurs. Les communications opportunistes permettent une couverture étendue dans les endroits où il n'existe aucune infrastructure réseau disponible et des stratégies de délestage de données pour aider les opérateurs à soulager la charge de leurs infrastructures. Dans cette thèse, nous considérons le cas de la diffusion opportuniste de plusieurs grands contenus d'un point de vue expérimental. Dans la première partie nous commençons par implémenter EPICS, un protocole réseau conçu pour l'échange opportuniste de grands contenus, dans des terminaux Android. Après sa évaluation nous proposons DAD, un nouveau protocole, qui envoie une rafale de paquets de données de façon adaptative. Nous comparons les deux protocoles expérimentalement et, à l'aide des traces de contacts, soit réelles, soit synthétiques, nous obtenons des gains importants avec cette nouvelle approche. La deuxième partie est dédiée au passage à l'échelle des systèmes de surveillance passive. Nous proposons deux approches. La première est basée sur la similarité des traces et des algorithmes de détection de communautés. La deuxième est basée sur des mesures collaboratives.