Blockchain for federated learning toward secure distributed machine learning systems: a systemic survey.

Federated learning () is a promising decentralized deep learning technology, which allows users to update models cooperatively without sharing their data. is reshaping existing industry paradigms for mathematical modeling and analysis, enabling an increasing number of industries to build privacy-preserving, secure distributed machine learning models. However, the inherent characteristics of have led to problems such as privacy protection, communication cost, systems heterogeneity, and unreliability model upload in actual operation.