arXiv · 2005.10785
Stochastic Optimization with Heavy-Tailed Noise via Accelerated Gradient Clipping
Abstract
In this paper, we propose a new accelerated stochastic first-order method called clipped-SSTM for smooth convex stochastic optimization with heavy-tailed distributed noise in stochastic gradients and derive the first high-probability complexity bounds for this method closing the gap in the theory of stochastic optimization with heavy-tailed noise. Our method is based on a special variant of accelerated Stochastic Gradient Descent (SGD) and clipping of stochastic gradients. We extend our method to the strongly convex case and prove new complexity bounds that outperform state-of-the-art results in this case. Finally, we extend our proof technique and derive the first non-trivial high-probability complexity bounds for SGD with clipping without light-tails assumption on the noise.
Explore related subjects
Keep this discovery
Eduard Gorbunov, Marina Danilova, Alexander Gasnikov. 2020-05-21. Stochastic Optimization with Heavy-Tailed Noise via Accelerated Gradient Clipping. https://arxiv.org/abs/2005.10785
Cite the original work for its findings. Save a collection to share your selection of sources.