{"id":"arxiv/2605.15314","url":"https://arxiv.org/abs/2605.15314","tags":["cs.LG","math.OC"],"score":0,"title":"Beyond Bounded Variance: Variance-Reduced Normalized Methods for Nonconvex Optimization under Blum-Gladyshev Noise","domain":"cs.LG","source":"arxiv","authors":["Antesh Upadhyay","Arda Fazla","Abolfazl Hashemi"],"reviews":0,"subject":"cs.LG","summary":"We study nonconvex stochastic optimization under the Blum-Gladyshev ($\\mathsf{BG}$-0) noise model, where the stochastic gradient variance grows quadratically with the distance from the initialization. We consider this problem under both standard smoothness and the symmetric generalized-smoothness framework, which captures objectives whose local curvature can scale with the gradient norm. We prove that normalized stochastic gradient descent with momentum, using only one stochastic gradient per iteration, converges under $\\mathsf{BG}$-0 noise with oracle complexity $O(\\varepsilon^{-6})$. This…","category":"research","comments":0,"disputed":false,"division":"cs","mentions":0,"discussed":0,"published":"2026-05-14","reactions":{}}