arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~
arXiv 2102.06489math.OCcs.LG
收藏

随机梯度裁剪的稳定性和收敛性:超越Lipschitz连续性和光滑性

Stability and Convergence of Stochastic Gradient Clipping: Beyond Lipschitz Continuity and Smoothness

KTH Royal Institute of Technology(瑞典皇家理工学院)

Vien V. Mai, Mikael Johansson

摘要:

随机梯度算法在应用于不具有Lipschitz连续和/或有界梯度的函数时通常是不稳定的。梯度裁剪是一种简单而有效的技术,可以稳定训练过程,解决容易出现梯度爆炸问题的问题。尽管它的广泛流行,梯度裁剪启发式算法的收敛特性知之甚少,特别是对于随机问题。建立了次梯度快速增长的非光滑凸函数的剪裁随机(次)梯度法(SCD)的定性和定量收敛结果。我们的分析表明,裁剪增强了SGD的稳定性,裁剪SGD算法在许多情况下具有有限的收敛速度。在标准假设下,我们还研究了带动量的剪切方法的收敛性,其中包括剪切SGD作为特殊情况。与一个新的李雅普诺夫分析,我们表明,所提出的方法达到了最好的已知率所考虑的一类问题,证明了有效性的裁剪方法也在这一制度。数值结果证实了我们的理论发展。

英文摘要:

Stochastic gradient algorithms are often unstable when applied to functions that do not have Lipschitz-continuous and/or bounded gradients. Gradient clipping is a simple and effective technique to stabilize the training process for problems that are prone to the exploding gradient problem. Despite its widespread popularity, the convergence properties of the gradient clipping heuristic are poorly understood, especially for stochastic problems. This paper establishes both qualitative and quantitative convergence results of the clipped stochastic (sub)gradient method (SGD) for non-smooth convex functions with rapidly growing subgradients. Our analyses show that clipping enhances the stability of SGD and that the clipped SGD algorithm enjoys finite convergence rates in many cases. We also study the convergence of a clipped method with momentum, which includes clipped SGD as a special case, for weakly convex problems under standard assumptions. With a novel Lyapunov analysis, we show that the proposed method achieves the best-known rate for the considered class of problems, demonstrating the effectiveness of clipped methods also in this regime. Numerical results confirm our theoretical developments.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑