Skip to main content

Privacy-preserving federated machine learning in ATM: experimental results from two use cases

Paper ID

SIDs-2022-104

Conference

SESAR Innovation Days

Year

2022

Theme

Machine Learning II

Project Name

SESAR 2020 ER4 project AICHAIN

Keywords:

Air Traffic Management, Demand and Capacity Balancing, federated learning, privacy-preserving machine learning

Authors

Sergio Ruiz, Javier Busto, Ignacio Martín, Salman Toor and Andre Rungger

DOI

Project Number

894162

Abstract

This paper presents the experimental results conducted with a new technological solution that can enable the privacy-preserving exploitation of large private datasets through collaborative machine learning. The solution has been developed under the exploratory research project AICHAIN (SESAR2020 ER04). The final aim of this solution is to enable air traffic management (ATM) operations to be improved with the added value provided by datasets that may be subject to strict privacy requirements and cannot be shared. Experiments have been run with two relevant use cases around Demand Capacity Balancing (DCB) services. Results prove that airline´s private data can improve the machine learning models performance in operations.