Saved in:
Bibliographic Details
Main Authors: Zhao, Shen-Yi, Shi, Chang-Wei, Xie, Yin-Peng, Li, Wu-Jun
Format: Preprint
Published: 2020
Subjects:
Online Access:https://arxiv.org/abs/2007.13985
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866929312584695808
author Zhao, Shen-Yi
Shi, Chang-Wei
Xie, Yin-Peng
Li, Wu-Jun
author_facet Zhao, Shen-Yi
Shi, Chang-Wei
Xie, Yin-Peng
Li, Wu-Jun
contents Stochastic gradient descent~(SGD) and its variants have been the dominating optimization methods in machine learning. Compared to SGD with small-batch training, SGD with large-batch training can better utilize the computational power of current multi-core systems such as graphics processing units~(GPUs) and can reduce the number of communication rounds in distributed training settings. Thus, SGD with large-batch training has attracted considerable attention. However, existing empirical results showed that large-batch training typically leads to a drop in generalization accuracy. Hence, how to guarantee the generalization ability in large-batch training becomes a challenging task. In this paper, we propose a simple yet effective method, called stochastic normalized gradient descent with momentum~(SNGM), for large-batch training. We prove that with the same number of gradient computations, SNGM can adopt a larger batch size than momentum SGD~(MSGD), which is one of the most widely used variants of SGD, to converge to an $ε$-stationary point. Empirical results on deep learning verify that when adopting the same large batch size, SNGM can achieve better test accuracy than MSGD and other state-of-the-art large-batch training methods.
format Preprint
id arxiv_https___arxiv_org_abs_2007_13985
institution arXiv
publishDate 2020
record_format arxiv
spellingShingle Stochastic Normalized Gradient Descent with Momentum for Large-Batch Training
Zhao, Shen-Yi
Shi, Chang-Wei
Xie, Yin-Peng
Li, Wu-Jun
Machine Learning
Stochastic gradient descent~(SGD) and its variants have been the dominating optimization methods in machine learning. Compared to SGD with small-batch training, SGD with large-batch training can better utilize the computational power of current multi-core systems such as graphics processing units~(GPUs) and can reduce the number of communication rounds in distributed training settings. Thus, SGD with large-batch training has attracted considerable attention. However, existing empirical results showed that large-batch training typically leads to a drop in generalization accuracy. Hence, how to guarantee the generalization ability in large-batch training becomes a challenging task. In this paper, we propose a simple yet effective method, called stochastic normalized gradient descent with momentum~(SNGM), for large-batch training. We prove that with the same number of gradient computations, SNGM can adopt a larger batch size than momentum SGD~(MSGD), which is one of the most widely used variants of SGD, to converge to an $ε$-stationary point. Empirical results on deep learning verify that when adopting the same large batch size, SNGM can achieve better test accuracy than MSGD and other state-of-the-art large-batch training methods.
title Stochastic Normalized Gradient Descent with Momentum for Large-Batch Training
topic Machine Learning
url https://arxiv.org/abs/2007.13985