SCALA: Split Federated Learning with Concatenated Activations and Logit Adjustments

TL;DR

SCALA introduces a novel split federated learning approach that concatenates client activations and adjusts logits to effectively handle label distribution skew caused by data heterogeneity and partial participation.

Contribution

It proposes a new method combining concatenated activations and logit adjustments to improve federated learning under skewed data distributions.

Findings

SCALA outperforms existing methods on public datasets.

Theoretical analysis confirms its robustness against label skew.

Experimental results demonstrate improved accuracy and convergence.

Abstract

Split Federated Learning (SFL) is a distributed machine learning framework which strategically divides the learning process between a server and clients and collaboratively trains a shared model by aggregating local models updated based on data from distributed clients. However, data heterogeneity and partial client participation result in label distribution skew, which severely degrades the learning performance. To address this issue, we propose SFL with Concatenated Activations and Logit Adjustments (SCALA). Specifically, the activations from the client-side models are concatenated as the input of the server-side model so as to centrally adjust label distribution across different clients, and logit adjustments of loss functions on both server-side and client-side models are performed to deal with the label distribution variation across different subsets of participating clients. Theoretical analysis and experimental results verify the superiority of the proposed SCALA on public datasets.