Federated Learning


Federated learning lets multiple devices or organizations train a shared model while keeping their raw training data local. Each participant performs local computation and exchanges model information with a coordinating server, making communication and data differences central challenges. My research examines how local updates affect generalization and how federated split learning can train more efficiently.

System Diagram

Federated learning system with local client data, local training, model update aggregation, and redistribution of a shared model.
Clients train on their own data and exchange model updates with a coordinator. Repeated aggregation and redistribution improve a shared model while raw data remain at each client.

My publications on this area

Peer-Reviewed Conference Papers

  1. Srijith Nair, Michael Lin, Peizhong Ju, Amirreza Talebi, Elizabeth Serena Bentley, and Jia Liu, "FSL-SAGE: Accelerating Federated Split Learning via Smashed Activation Gradient Estimation," 42nd International Conference on Machine Learning (ICML'25). arXiv version ICML
    @inproceedings{nair2025fsl,
      title={FSL-SAGE: Accelerating Federated Split Learning via Smashed Activation Gradient Estimation},
      author={Nair, Srijith and Lin, Michael and Ju, Peizhong and Talebi, Amirreza and Bentley, Elizabeth and Liu, Jia},
      booktitle={International Conference on Machine Learning},
      year={2025},
      organization={PMLR}
    }
    
  2. Peizhong Ju, Haibo Yang, Jia Liu, Yingbin Liang, and Ness B. Shroff, "Can We Theoretically Quantify the Impacts of Local Updates on the Generalization Performance of Federated Learning?" 25th International Symposium on Theory, Algorithmic Foundations, and Protocol Design for Mobile Networks and Mobile Computing (MobiHoc'24), 2024 . arXiv version
    @inproceedings{ju2024can,
      title={Can We Theoretically Quantify the Impacts of Local Updates on the Generalization Performance of Federated Learning?},
      author={Ju, Peizhong and Yang, Haibo and Liu, Jia and Liang, Yingbin and Shroff, Ness},
      booktitle={Proceedings of the Twenty-fifth International Symposium on Theory, Algorithmic Foundations, and Protocol Design for Mobile Networks and Mobile Computing (MobiHoc'24)},
      pages={141--150},
      year={2024}
    }