Decentralized Federated Averaging
About
Federated averaging (FedAvg) is a communication efficient algorithm for the distributed training with an enormous number of clients. In FedAvg, clients keep their data locally for privacy protection; a central parameter server is used to communicate between clients. This central server distributes the parameters to each client and collects the updated parameters from clients. FedAvg is mostly studied in centralized fashions, which requires massive communication between server and clients in each communication. Moreover, attacking the central server can break the whole system's privacy. In this paper, we study the decentralized FedAvg with momentum (DFedAvgM), which is implemented on clients that are connected by an undirected graph. In DFedAvgM, all clients perform stochastic gradient descent with momentum and communicate with their neighbors only. To further reduce the communication cost, we also consider the quantized DFedAvgM. We prove convergence of the (quantized) DFedAvgM under trivial assumptions; the convergence rate can be improved when the loss function satisfies the P{\L} property. Finally, we numerically verify the efficacy of DFedAvgM.
Related benchmarks
| Task | Dataset | Result | Rank | |
|---|---|---|---|---|
| Image Classification | Tiny-ImageNet Dirichlet alpha=0.1 (test) | Test Accuracy24.42 | 30 | |
| Image Classification | Cifar10 Dirichlet(0.3) (test) | Test Accuracy82.7 | 21 | |
| Image Classification | Tiny-ImageNet Dirichlet alpha=0.3 (test) | Test Accuracy16.51 | 10 | |
| Image Classification | Tiny-ImageNet Pathological c=20 (test) | Test Accuracy31.5 | 10 | |
| Image Classification | Tiny-ImageNet Pathological c=10 (test) | Test Accuracy41.94 | 10 | |
| Image Classification | CIFAR-10 (test) | Accuracy (Dirichlet α=0.1)86.94 | 10 | |
| Image Classification | CIFAR-10 Pathological-2 (test) | Test Accuracy91.52 | 9 | |
| Personalized Federated Learning | Tiny-ImageNet Pathological 10 | Accuracy@40173 | 9 | |
| Personalized Federated Learning | Tiny-ImageNet Pathological 20 | Top-30 Accuracy210 | 8 | |
| Image Classification | CIFAR-100 (Pat-10) | Communication Rounds (acc@65)249 | 8 |