{"ID":2841974,"CreatedAt":"2026-06-01T04:54:23.091178241Z","UpdatedAt":"2026-06-01T04:54:23.091178241Z","DeletedAt":null,"paper_url":"https://arxiv.org/abs/2511.09828","arxiv_id":"2511.09828","title":"SMoFi: Step-wise Momentum Fusion for Split Federated Learning on Heterogeneous Data","abstract":"Split Federated Learning is a system-efficient federated learning paradigm that leverages the rich computing resources at a central server to train model partitions. Data heterogeneity across silos, however, presents a major challenge undermining the convergence speed and accuracy of the global model. This paper introduces Step-wise Momentum Fusion (SMoFi), an effective and lightweight framework that counteracts gradient divergence arising from data heterogeneity by synchronizing the momentum buffers across server-side optimizers. To control gradient divergence over the training process, we design a staleness-aware alignment mechanism that imposes constraints on gradient updates of the server-side submodel at each optimization step. Extensive validations on multiple real-world datasets show that SMoFi consistently improves global model accuracy (up to 7.1%) and convergence speed (up to 10.25$\\times$). Furthermore, SMoFi has a greater impact with more clients involved and deeper learning models, making it particularly suitable for model training in resource-constrained contexts.","short_abstract":"Split Federated Learning is a system-efficient federated learning paradigm that leverages the rich computing resources at a central server to train model partitions. Data heterogeneity across silos, however, presents a major challenge undermining the convergence speed and accuracy of the global model. This paper introd...","url_abs":"https://arxiv.org/abs/2511.09828","url_pdf":"https://arxiv.org/pdf/2511.09828v2","authors":"[\"Mingkun Yang\",\"Ran Zhu\",\"Qing Wang\",\"Jie Yang\"]","published":"2025-11-13T00:21:05Z","proceeding":"cs.LG","tasks":"[\"cs.LG\",\"cs.DC\"]","methods":"[]","has_code":false}