English

Clipped Stochastic Methods for Variational Inequalities with Heavy-Tailed Noise

Optimization and Control 2022-11-02 v2 Machine Learning

Abstract

Stochastic first-order methods such as Stochastic Extragradient (SEG) or Stochastic Gradient Descent-Ascent (SGDA) for solving smooth minimax problems and, more generally, variational inequality problems (VIP) have been gaining a lot of attention in recent years due to the growing popularity of adversarial formulations in machine learning. However, while high-probability convergence bounds are known to reflect the actual behavior of stochastic methods more accurately, most convergence results are provided in expectation. Moreover, the only known high-probability complexity results have been derived under restrictive sub-Gaussian (light-tailed) noise and bounded domain assumption [Juditsky et al., 2011]. In this work, we prove the first high-probability complexity results with logarithmic dependence on the confidence level for stochastic methods for solving monotone and structured non-monotone VIPs with non-sub-Gaussian (heavy-tailed) noise and unbounded domains. In the monotone case, our results match the best-known ones in the light-tails case [Juditsky et al., 2011], and are novel for structured non-monotone problems such as negative comonotone, quasi-strongly monotone, and/or star-cocoercive ones. We achieve these results by studying SEG and SGDA with clipping. In addition, we numerically validate that the gradient noise of many practical GAN formulations is heavy-tailed and show that clipping improves the performance of SEG/SGDA.

Keywords

Cite

@article{arxiv.2206.01095,
  title  = {Clipped Stochastic Methods for Variational Inequalities with Heavy-Tailed Noise},
  author = {Eduard Gorbunov and Marina Danilova and David Dobre and Pavel Dvurechensky and Alexander Gasnikov and Gauthier Gidel},
  journal= {arXiv preprint arXiv:2206.01095},
  year   = {2022}
}

Comments

NeurIPS 2022. 74 pages, 18 figures. Changes in v2: few typos were fixed, new experiments with clipped-SEG were added. Code: https://github.com/busycalibrating/clipped-stochastic-methods

R2 v1 2026-06-24T11:37:18.916Z