Training Latency Minimization for Model-Splitting Allowed Federated Edge Learning
Yao Wen, Guopeng Zhang, Kezhi Wang, Kun Yang · IEEE Transactions on Network Science and Engineering · 2025
To alleviate the shortage of computing power faced by clients in training deep neural networks (DNNs) using federated learning (FL), we leverage theedge computingandsplit learningto propose a model-splitting allowed FL (SFL) framework, with the aim to minimize the training latency without loss of test accuracy. Under thesynchronized global updatesetting, the latency to complete a round of global training is determined by the maximum latency for the clients to complete a local training session. Therefore, the training latency minimization problem (TLMP) is modelled as a minimizing-maximum problem. To solve this mixed integer nonlinear programming problem, we first propose aregression methodto fit the quantitative-relationship between thecut-layerand other parameters of an AI-model, and thus, transform the TLMP into a continuous problem. Considering that the two subproblems involved in the TLMP, namely, thecut-layer selection problemfor the clients and thecomputing resource allocation problemfor the parameter-server are relative independence, an alternate-optimization-based algorithm with polynomial time complexity is developed to obtain a high-quality solution to the TLMP. Extensive experiments are performed on a popular DNN-modelEfficientNetV2using dataset MNIST, and the results verify the validity and improved performance of the proposedSFLframework.