Timezone: »

Tackling the Objective Inconsistency Problem in Heterogeneous Federated Optimization
Jianyu Wang · Qinghua Liu · Hao Liang · Gauri Joshi · H. Vincent Poor

Thu Dec 10 09:00 AM -- 11:00 AM (PST) @ Poster Session 5 #1636

In federated learning, heterogeneity in the clients' local datasets and computation speeds results in large variations in the number of local updates performed by each client in each communication round. Naive weighted aggregation of such models causes objective inconsistency, that is, the global model converges to a stationary point of a mismatched objective function which can be arbitrarily different from the true objective. This paper provides a general framework to analyze the convergence of federated heterogeneous optimization algorithms. It subsumes previously proposed methods such as FedAvg and FedProx and provides the first principled understanding of the solution bias and the convergence slowdown due to objective inconsistency. Using insights from this analysis, we propose FedNova, a normalized averaging method that eliminates objective inconsistency while preserving fast error convergence.

Author Information

Jianyu Wang (Carnegie Mellon University)
Qinghua Liu (Princeton University)
Hao Liang (Carnegie Mellon University)
Gauri Joshi (Carnegie Mellon University)
H. Vincent Poor (Princeton University)

More from the Same Authors