Browse Publications Technical Papers 2020-01-0139
2020-04-14

A Secure and Privacy-Preserving Collaborative Machine Learning System for Intelligent Transportation System 2020-01-0139

Modern vehicles are increasingly equipped with multiple advanced on-board sensors and keep generating large volumes of data. Along with the recent advances in a wide range of Machine Learning (ML) algorithms, the vehicular data are being analyzed intelligently to enable users to be better informed and make safer, more coordinated, and smarter use of transport networks. The success of ML model relies on the availability of large set of relevant data so that the underlying model can be trained better. However, it is not possible for a ML model to fetch the complete set of data from a single vehicle, thus, the collaboration of other vehicles are desired in sharing their local model and collaboratively training the model. Collaborative machine learning (CML) mechanism can improve the intelligence of the ML models in different vehicles by transferring the learned knowledge from the local ML model of one vehicle to another across the distributed network. However, the privacy concerns related to sharing the ML models often create hindrance towards sharing the local model with others. Sharing the information related to the local model may also leak sensitive data/patterns pertaining to the vehicles user. The aim of the paper is to discuss the privacy issues specific to sharing ML model in the collaborative ML scenarios, and propose the use of group signature scheme to enable security and privacy. The security and performance of the proposed system based on the group signature scheme is evaluated for the intelligent transportation systems.

SAE MOBILUS

Subscribers can view annotate, and download all of SAE's content. Learn More »

Access SAE MOBILUS »

Members save up to 18% off list price.
Login to see discount.
Special Offer: Download multiple Technical Papers each year? TechSelect is a cost-effective subscription option to select and download 12-100 full-text Technical Papers per year. Find more information here.
X