FedOUI: OUI-Guided Client Weighting for Federated Aggregation

TL;DR

FedOUI introduces an activation-based, label-free metric for client weighting in federated learning, improving aggregation especially under high heterogeneity by leveraging internal model signals.

Contribution

The paper proposes FedOUI, a novel aggregation rule using the Overfitting-Underfitting Indicator to better weight clients based on internal activation signals.

Findings

FedOUI outperforms FedAvg, FedProx, and gradient-alignment baselines under non-IID and noisy conditions.

OUI-based weighting improves model aggregation quality in heterogeneous federated settings.

The method remains lightweight and interpretable, utilizing internal activation structures.

Abstract

Federated learning usually aggregates client updates using dataset size or gradient-level criteria, while overlooking internal signals about how each client model is organizing its input space during training. We introduce FedOUI, a simple aggregation rule based on the Overfitting-Underfitting Indicator (OUI), an activation-based and label-free metric. Each participating client sends its local update together with a OUI value computed on a fixed probe batch, and the server estimates the round-wise OUI distribution to assign lower weights to structurally atypical clients through a smooth reweighting rule. We evaluate FedOUI on CIFAR-10 under strong non-IID partitioning and noisy-client conditions, comparing it with FedAvg, FedProx, and a gradient-alignment baseline. The clearest gains appear under strong heterogeneity, where OUI-based weighting improves aggregation quality while remaining lightweight and interpretable. These results show that internal activation structure can provide useful information for federated aggregation beyond client size and gradient geometry.