arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2103.13445cs.LG
收藏

一种简单有效的低精度训练神经网络的随机舍入方法

A Simple and Efficient Stochastic Rounding Method for Training Neural Networks in Low Precision

Eindhoven University of Technology(埃因霍温理工大学)

Lu Xia, Martijn Anthonissen, Michiel Hochstenbach, Barry Koren

摘要:

传统的随机舍入(CSR)被广泛应用于神经网络(NN)的训练中,即使在低精度计算中也显示出良好的训练结果。本文介绍了一种改进的随机舍入方法,该方法简单有效。所提出的方法成功地训练神经网络与16位定点数,并提供更快的收敛速度和更高的分类精度比CSR和确定性舍入到最近的方法。

英文摘要:

Conventional stochastic rounding (CSR) is widely employed in the training of neural networks (NNs), showing promising training results even in low-precision computations. We introduce an improved stochastic rounding method, that is simple and efficient. The proposed method succeeds in training NNs with 16-bit fixed-point numbers and provides faster convergence and higher classification accuracy than both CSR and deterministic rounding-to-the-nearest method.

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。

↑