- 期刊
Research on Federated Learning for Non‐IID Data
摘要
In recent years, federated learning has attracted widespread attention as a technology to solve the problem of data islands, and has begun to be applied in fields such as finance, healthcare, and smart cities. Introduce federated learning from three levels. First introduce the concept of federated learning, and explain the concept of federated learning through the definition, architecture, classification and comparison with traditional distributed learning; then from the perspective of machine learning and deep learning, various current federated learning algorithms are analyzed Classification comparison and in‐depth analysis; finally, in‐depth classification of federated learning optimization algorithms from the perspectives of communication cost, client selection, and aggregation mode optimization, summarizes the research status of federated learning, and puts forward the communication and system differences faced by federated learning. Data heterogeneity and data heterogeneity are the three major problems and solutions, as well as the outlook for the future.