Track papers across labs, topics, and trends.
Search papers, labs, and topics across Lattice.
This paper investigates streaming federated learning under constraints of limited client memory and variable sampling costs, proposing a joint server-side admission and client-side memory management framework aimed at minimizing cumulative excess population risk. The authors derive a learning-error bound that accounts for factors like training sample size and reuse imbalance, leading to the development of the Active-Constraint Drift-Plus-Penalty (ACDPP) policy, which effectively balances data retention and admission strategies. Experimental results show that ACDPP closely approximates an oracle benchmark while adhering to sampling-cost and buffer constraints, demonstrating its practical applicability in real-world scenarios.
ACDPP policy achieves near-oracle performance in streaming federated learning while managing memory and sampling costs effectively.
We study streaming federated learning with limited client memory, where newly generated training data incur time-varying sampling costs and must be selectively admitted and retained over time. We consider a joint server-side admission and client-side memory-management framework with the objective of minimizing the cumulative excess population risk under a sampling-cost budget and buffer constraints. We first derive a learning-error bound that explicitly captures the effects of instantaneous training sample size, distinct-sample growth, and reuse imbalance through a characterization of the effective sample size. Through a surrogate penalty obtained from this bound, we develop an Active-Constraint Drift-Plus-Penalty (ACDPP) policy that combines a structured client-side $K$-step retention rule with a server-side online admission rule and a time-varying rectangular admission region. We further present a sequence of comparison arguments, via an auxiliary constant-admission policy, that connects the ACDPP learning bound to a costless oracle benchmark. This yields explicit guarantees in terms of sublinear regret and sampling-cost violation, while the buffer-occupancy violation is controlled through offline selection of the retention horizon. Experiments on multiple datasets demonstrate that the proposed policy remains close to the oracle benchmark while satisfying the sampling-cost and buffer constraints.