Convergence Analysis of Sequential Split Learning on Heterogeneous Data
This provides theoretical justification for using SSL over FedAvg in distributed machine learning with heterogeneous data, which is incremental but addresses a known bottleneck.
The paper tackles the lack of rigorous convergence analysis for Sequential Split Learning (SSL) by deriving convergence guarantees for strongly convex, general convex, and non-convex objectives on heterogeneous data, showing that SSL outperforms Federated Averaging (FedAvg) in such settings, with empirical validation on extremely heterogeneous data.
Federated Learning (FL) and Split Learning (SL) are two popular paradigms of distributed machine learning. By offloading the computation-intensive portions to the server, SL is promising for deep model training on resource-constrained devices, yet still lacking of rigorous convergence analysis. In this paper, we derive the convergence guarantees of Sequential SL (SSL, the vanilla case of SL that conducts the model training in sequence) for strongly/general/non-convex objectives on heterogeneous data. Notably, the derived guarantees suggest that SSL is better than Federated Averaging (FedAvg, the most popular algorithm in FL) on heterogeneous data. We validate the counterintuitive analysis result empirically on extremely heterogeneous data.