Introduction
The growth and divergence of regulations governing the sharing of data across jurisdictions are an increasing fact of life for international organisations.
HSBC would like to examine the use of techniques such as homomorphic encryption (HE), SME and Federated Learning/Analytics to remove the need to share personal data across jurisdictions, allowing them to more easily comply with their regulatory obligations.
Explaining the science
In this scenario there is a trade-off with accuracy that can be managed and reduced through an improved understanding of the available methodology. By reducing the need to move data cross-border, they will also be able to work more effectively with key and target markets, that may have local data sharing restrictions in place.
Project aims
Outputs of this project will include:
- Analysing the state of the art in federated learning, federated analytics and trustworthy information sharing.
- Analysing the data to clarify the advantages and disadvantages associated with each tool of approach.
- Providing HSBC with a menu of potential PET solutions and process flow documentation to support understanding of how the code can be used and the different parameters.
- Creating expert guidance notes and a white paper. These will support HSBC to navigate the offer of potential providers across Technology, Concept and Research.
- Develop, Analyse and Test algorithms for Federated Analytics based on secure multi-party computation (SMC) and homomorphic encryption (HE).
- Work with HSBC to identify further needs and use cases in PETS (including federated analytics) and learning.
Applications
As a leader in financing global trade, this technology is seen as an enabler for HSBC to continue to grow its business.
Organisers
Collaborators
