{"id":"arxiv/2601.20738","url":"https://arxiv.org/abs/2601.20738","tags":["cs.LG","cs.DC","eess.SP","math.OC","stat.ML"],"score":0,"title":"SA-PEF: Step-Ahead Partial Error Feedback for Efficient Federated Learning","domain":"cs.LG","source":"arxiv","authors":["Dawit Kiros Redie","Reza Arablouei","Stefan Werner"],"reviews":0,"subject":"cs.LG","summary":"Biased gradient compression with error feedback (EF) reduces communication in federated learning (FL), but under non-IID data, the residual error can decay slowly, causing gradient mismatch and stalled progress in the early rounds. We propose step-ahead partial error feedback (SA-PEF), which integrates step-ahead (SA) correction with partial error feedback (PEF). SA-PEF recovers EF when the step-ahead coefficient $α=0$ and step-ahead EF (SAEF) when $α=1$. For non-convex objectives and $δ$-contractive compressors, we establish a second-moment bound and a residual recursion that guarantee…","category":"research","comments":0,"disputed":false,"division":"cs","mentions":0,"discussed":0,"published":"2026-01-28","reactions":{}}