Federated Learning for Credit Scoring Model Using Blockchain

In the last years technologies have emerged in a very disruptive way, modelling and navigating a big part of human life. Recently blockchain and machine learning technologies reached a new level of maturity. From low penetrated tech, blockchain become in the bottom of the digital crypto currencies and many projects out of them. At the same time, machine learning is becoming more powerful and strong technology and many companies start to adopt it as a solution for non-trivial technical problems leveraging on the tones of data they use and generate in their daily functioning. Both blockchain and machine learning technologies have become a big part of our life and different studies show how much potential they have. The paper presents a system architecture approach for building a decentralized FELScore platform based on federated learning paradigm and blockchain technology for credit score modelling. The main goal is to provide a platform for distributed machine learning that several financial institutions could leverage on without having the need to exchange real customer or product data and without the requirement to trust each other. To further improve the security of the federated learning the proposed approach uses blockchain technology.