FedSysID: A Federated Approach to Sample-Efficient System Identification

This repository includes the MATLAB codes to implement the experimental results of the following paper:

1. H. Wang, L. F. Toso, J. Anderson (2022). FedSysID: A Federated Approach to Sample-Efficient System Identification

Instructions

To run the codes in this repository, you only need a working MATLAB installation. In this paper we implement our FedSysId algorithm with two different FL solvers:

• FedAvg - Ref [1]
• FedLin - Ref [2]

The nominal system considered in our experiments was first presented in Ref [3]

References

1. Brendan McMahan, Eider Moore, Daniel Ramage, Seth Hampson, Blaise Aguera y Arcas (2017) Communication-Efficient Learning of Deep Networks from Decentralized Data
2. Aritra Mitra, Rayana Jaafar, George J. Pappas, Hamed Hassani (2021) Linear Convergence in Federated Learning: Tackling Client Heterogeneity and Sparse Gradients
3. Lei Xin, Lintao Ye, George Chiu, Shreyas Sundaram (2022) Identifying the Dynamics of a System by Leveraging Data from Similar Systems

Troubleshooting

If you have any trouble running those codes or have any question about the paper, please email Leonardo F. Toso or Han Wang.