{"id":"arxiv/2606.22466","url":"https://arxiv.org/abs/2606.22466","tags":["cs.LG","math.OC"],"score":0,"title":"Federated learning with heavy-tailed gradient noise and communication noise: a variance-reduction based algorithm","domain":"cs.LG","source":"arxiv","authors":["Shengchao Zhao","Yongchao Liu"],"reviews":0,"subject":"cs.LG","summary":"Federated learning (FL) is an emerging distributed machine learning paradigm that enables local devices to jointly train a global model while keeping data decentralized and private. We propose a variance-reduction based algorithm, VRA-FedSGD, for FL in the presence of heavy-tailed gradient noise and communication noise, where these noises are prevalent in large-scale machine learning over wireless networks and Internet of Things deployments. VRA-FedSGD employs a momentum variance reduction technique together with a nonlinear mapping to mitigate heavy-tailed gradient noise, and uses a…","category":"research","comments":0,"disputed":false,"division":"cs","mentions":0,"discussed":0,"published":"2026-06-21","reactions":{}}