Open Banking pioneer LUXHUB announced a strategic partnership with the University of Luxembourg’s Interdisciplinary Centre for Security, Reliability and Trust (SnT) to create added value  services based on financial data. The partnership will implement ground-breaking technology in the field of artificial intelligence while respecting the industry’s main concerns - the safety and privacy of sensitive data – through the federated learning model.  

“In the era of Open Finance – with financial service companies collaborating, co-creating  new products and services by taking advantage of the API economy – it makes sense for  them to opt for the federated learning model, and keep building the future of finance. Together,” first explains Claude Meurisse, COO of LUXHUB. 

Federated learning consists of a machine learning procedure, where the goal is to train a  high-quality model with data distributed over several independent providers, such that the  data remains locked on each provider, and no centralized data storage is actually needed.

Federated learning differs from most machine learning models, as the algorithm – and  NOT the data – travels. Claude Meurisse comments: “Joint predictive models are shared and created by what is called the federation. And eventually, all participants learn from  everyone without learning from anyone.”

A secure and confidential learning model 

Through this federated learning model, LUXHUB and SnT are making sure data never  leaves the premises of the client/participant, and therefore security and confidentiality are  ensured.

“Federated learning allows the training of a model with your own data but without sharing  the data with anyone else; you just share updates to a global model. The data stays on  each client’s premises – in the case of a bank, down to a single branch – to avoid any risks,  but the system allows every participant to leverage the collective intelligence to train the  global model. It’s a win-win: everyone is better off by collaborating, while ensuring data  security and privacy at the same time,” highlights Prof. Radu State, Principal Investigator  of the project at SnT.

Moreover, leveraging federated learning in the financial sector counts more benefits, from  lower latency and less power consumption, to a significant decrease in false positives and therefore an important decrease in operating costs.

Key use cases within the financial sector 

LUXHUB, as a central and neutral piece of this data puzzle, aims to enable more  interaction and innovation within the financial services industry. “LUXHUB is all about  fostering innovation and collaboration,” underlines the COO.

More concretely, the participants will power an algorithm, making it more efficient and  accurate, allowing the financial institutions to take better-informed decisions. The training  of the algorithm will be assumed by the LUXHUB team in collaboration with SnT researchers.

Phase 1 of the project will consist of the design and management of a secure federated  learning platform. In parallel, the teams will be focusing on several key use-cases that will benefit the entire financial services industry, such as fraud detection, anti-money  laundering, loan risk prediction, and transaction categorisation. The model could then be  extended to additional use cases, dealing notably with key compliance topics, and more.

“Leveraging data and the knowledge of SnT researchers, this project with LUXHUB on  federated machine learning is highly innovative, aiming at identifying illicit financial activity  by enabling shared learning, but without any risk in sharing data. The project outcomes  have tremendous potential allowing the design of solutions that leverage data across  several stakeholders in a secure and compliant manner,” said Björn Ottersten, Director of  SnT.

Source: LUXHUB