With the growing computing power in the Internet of Vehicles (IoV), machine learning is increasingly utilized. Yet, IoV faces challenges like privacy, security, and trust issues between vehicles and infrastructure, hindering efficient information usage and machine learning. This paper introduces Semi-Supervised Federated Learning (SSFL) for object recognition in IoV. The proposed approach is designed to enhance the generalization capability and improve the performance of the algorithm by adapting the SSFL framework to the specific characteristics of IoV data deployment and algorithms. A teacher-student structured approach leverages labeled and unlabeled data, and a deployment scheme optimizes training. Results surpass traditional methods, promising improved IoV object recognition accuracy and efficiency.
An Enhancing Semi-Supervised Federated Learning Framework for Internet of Vehicles
10.10.2023
1876178 byte
Aufsatz (Konferenz)
Elektronische Ressource
Englisch
METHODS TO IMPROVE FEDERATED LEARNING ROBUSTNESS IN INTERNET OF VEHICLES
Europäisches Patentamt | 2024
|