arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 1901.09149cs.LGmath.OCstat.ML
收藏

用自适应梯度法摆脱鞍点

Escaping Saddle Points with Adaptive Gradient Methods

MIT EECS(麻省理工学院电子工程与计算机科学系);Google Research, New York(谷歌研究院纽约分部)

Matthew Staib, Sashank J. Reddi, Satyen Kale, Sanjiv Kumar, Suvrit Sra

摘要:

Adam和RMSProp等自适应方法广泛用于深度学习,但尚未得到很好的理解。在本文中,我们寻求一个清晰的,干净的和精确的表征他们的行为在非凸设置。为此,我们首先提供了一个新的自适应方法预处理SGD,其中预条件估计在线的方式。通过研究预处理器本身,我们阐明了它的目的:它重新调整随机梯度噪声在平稳点附近是各向同性的,这有助于逃离鞍点。此外,我们表明,自适应方法可以有效地估计上述预条件。通过粘合在一起,这两个组件,我们提供了第一(据我们所知)的二阶收敛结果的任何自适应方法。从我们的分析中得出的关键见解是,与SGD相比,自适应方法更快地逃离鞍点,并且可以更快地整体收敛到二阶稳定点。

英文摘要:

Adaptive methods such as Adam and RMSProp are widely used in deep learning but are not well understood. In this paper, we seek a crisp, clean and precise characterization of their behavior in nonconvex settings. To this end, we first provide a novel view of adaptive methods as preconditioned SGD, where the preconditioner is estimated in an online manner. By studying the preconditioner on its own, we elucidate its purpose: it rescales the stochastic gradient noise to be isotropic near stationary points, which helps escape saddle points. Furthermore, we show that adaptive methods can efficiently estimate the aforementioned preconditioner. By gluing together these two components, we provide the first (to our knowledge) second-order convergence result for any adaptive method. The key insight from our analysis is that, compared to SGD, adaptive methods escape saddle points faster, and can converge faster overall to second-order stationary points.

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。

↑