@inproceedings{dbff38564842447eaaae965195e012f3,
title = "Asynchronous Federated Learning Framework Based on Dynamic Selective Transmission",
abstract = "This paper proposes an asynchronous federated learning framework based on dynamic selective transmission to solve the communication efficiency problem in asynchronous federated learning. The framework first dynamically determines the network layer to be transmitted in each round according to the training progress, and then dynamically adjusts the ratio of parameters to be transmitted according to the training progress and model staleness for the selected network layers, which effectively reduces the negative impact of the local model staleness of the client on the performance of global model while significantly reducing the uplink communication cost of the asynchronous federated learning framework. On two datasets, this paper designs experiments to compare the proposed framework with the existing work. Through the analysis of the experimental results, the effectiveness of the proposed framework is verified.",
keywords = "Asynchronous federated learning, Communication efficiency, Federated learning, Machine learning",
author = "Ruizhuo Zhang and Wenjian Luo and Yongkang Luo and Shaocong Xue",
note = "Publisher Copyright: {\textcopyright} 2023, The Author(s), under exclusive license to Springer Nature Switzerland AG.; 14th International Conference on Advances in Swarm Intelligence, ICSI 2023 ; Conference date: 14-07-2023 Through 18-07-2023",
year = "2023",
doi = "10.1007/978-3-031-36625-3\_16",
language = "英语",
isbn = "9783031366246",
series = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",
publisher = "Springer Science and Business Media Deutschland GmbH",
pages = "193--203",
editor = "Ying Tan and Yuhui Shi and Wenjian Luo",
booktitle = "Advances in Swarm Intelligence - 14th International Conference, ICSI 2023, Proceedings",
address = "德国",
}