DG-FedReuse: Proxy-Gradient-Gated Cached-Update Reuse with Matched Sparse Uplink Accounting

2026-08-05

AI Summary

This work addresses the high communication overhead in federated learning caused by frequent local updates and model transmissions. The authors propose DG-FedReuse, a novel mechanism that introduces a gating strategy based on a proxy of head-gradient differences, enabling clients to reuse decayed cached updates when the proxy difference falls below a threshold. This approach is regulated by a fresh-client quota and a hard cache age limit. Additionally, adaptive tensor-level Top-K sparsification is employed to compress uplink transmissions. Under Dirichlet non-IID data partitions, DG-FedReuse reduces uplink communication by 83.36–85.42% over 90 training rounds compared to Top-K FedAvg, with only a 0.14–5.29 percentage point drop in accuracy. When accounting for symmetric downlink communication, total communication savings reach 41.68–42.71%.

Abstract

Federated learning repeatedly incurs local optimization and model-update transmission. We study DG-FedReuse, a simulator-level mechanism that allows selected clients to contribute age-decayed cached updates when a stochastic head-gradient discrepancy proxy remains below a round-dependent threshold. A hard cache-age limit and minimum fresh-client quota constrain reuse, while fresh updates use an adaptive per-tensor Top-K numerical-field representation. Experiments cover six image-classification datasets, 50 virtual clients, Dirichlet label heterogeneity (α=0.5), and three seeds. At a common 90-round budget, DG-FedReuse yields 83.36-85.42% modeled update-data-field uplink saving, compared with 76.88% for matched Top-K FedAvg; the seed-aligned accuracy differences range from -5.29 to -0.14 percentage points. Best-observed test accuracies obtained under test-controlled checkpointing are retained only as exploratory archival evidence and range from -2.38 to +0.45 percentage points relative to matched FedAvg. A symmetric dense-model-downlink sensitivity reduces the headline saving to 41.68-F42.71% and the incremental gain over Top-K FedAvg to 3.24-4.27 percentage points, demonstrating the dependence of communication conclusions on the accounting boundary. The study characterizes the proposed reuse rule in the implemented simulator; it does not establish unbiased generalization, end-to-end bandwidth reduction, runtime or energy savings, faster convergence, or superiority over existing stale-update and lazy-aggregation methods.

Problem

• federated learning
• communication efficiency
• update reuse
• uplink compression
• cached updates

Innovation

• cached-update reuse
• proxy-gradient gating
• adaptive Top-K sparsification
• federated learning communication efficiency
• stale update management