An Inclusive Finance Consortium Blockchain Platform for Secure Data Storage and Value Analysis for Small and Medium-Sized Enterprises

Abstract

In the era of big data, people pay more and more attention to user privacy and data security. The market size of Small and Medium-sized Enterprises (SMEs) in China is sizable. Nevertheless, due to problems of data dispersion and lack of data features, it is very difficult to make use of the massive data of SMEs scattered in various institutions effectively, which leads to the inability to reflect the value of data. One of the problems is how to credit for SMEs better. Supported by federated learning for such scenarios, we present an inclusive Finance Consortium Blockchain platform in this paper. On the one hand, the platform combines decentralized identity and Blockchain as the underlying architecture, which guarantees authenticity of the data source and safe storage of user data on the chain. The smart contract mechanism provided by Blockchain can also assist secure storage and incentive mechanism of federated learning models effectively. On the other hand, we have innovatively introduced an asynchronous Federated Learning mode based on transfer learning, which encrypts the well-designed pre-training model and transfers it to each participant to guide training of the participant's local model. In the model reasoning stage, all the participants participate in a joint evaluation according to the local model and push the reasoning results on the Blockchain. The smart contract takes the weighted sum of the reasoning results provided by all the participants as the final result of shared model reasoning.