Federated Imputation under Heterogeneous Feature Spaces

TL;DR

This paper introduces FedHF-Impute, a federated imputation method that leverages a global feature graph to enable knowledge transfer across clients with heterogeneous feature spaces, improving imputation accuracy.

Contribution

The paper proposes FedHF-Impute, a novel federated imputation framework that uses message passing on a global feature graph to handle non-overlapping feature spaces.

Findings

FedHF-Impute improves imputation RMSE by 26.9% on SECOM dataset.

It achieves 8.4% RMSE improvement on AirQuality dataset.

Performance is comparable to baselines on PhysioNET with minimal accuracy loss.

Abstract

Federated Learning (FL) enables collaborative training across decentralized clients, but most methods assume aligned feature schemas, an assumption that rarely holds in tabular settings where clients observe only partially overlapping feature subsets. In these heterogeneous feature spaces, parameter-averaging methods (e.g., FedAvg) transfer little information across weakly overlapping or disjoint feature groups, limiting their effectiveness for federated imputation. To overcome this, we propose \textbf{FedHF-Impute}, a federated imputation framework that separates structural feature unavailability from conventional missingness and uses a shared global feature graph to propagate information across statistically related features through message passing. This enables indirect cross-client knowledge transfer, even when features are never jointly observed locally, while preserving standard federated communication. Under simulated partial schema overlap on the SECOM and AirQuality datasets, FedHF-Impute improves imputation accuracy (RMSE) over FL baselines by 26.9\%, and 8.4\% respectively, while achieving comparable performance on PhysioNET, with only a 0.3\% difference relative to the best baseline.