arXivDaily arXiv每日学术速递 周一至周五更新
北美退伍军人节(Veterans Day),12 日无更新
arXiv 1802.06502cs.LG
收藏

EA-CG:一种训练全连接神经网络的近似二阶方法

EA-CG: An Approximate Second-Order Method for Training Fully-Connected Neural Networks

HTC Research & Healthcare(HTC研究与医疗)

Sheng-Wei Chen, Chun-Nan Chou, Edward Y. Chang

摘要:

为了训练全连接神经网络(FCNN),提出了一种实用的近似二阶方法,包括:1)Hessian矩阵的近似和2)基于共轭梯度(CG)的方法。我们提出的近似Hessian矩阵是记忆有效的,可以应用于任何FCNN,其中激活和标准函数是两次可微的。我们设计了一种基于CG的方法,结合一阶近似来推导训练FCNN的牛顿方向,这显着降低了空间和时间复杂度。这种基于CG的方法可以用来解决任何线性方程的系数矩阵是克罗内克因子,对称和正定。实验结果表明了该方法的有效性和有效性。

英文摘要:

For training fully-connected neural networks (FCNNs), we propose a practical approximate second-order method including: 1) an approximation of the Hessian matrix and 2) a conjugate gradient (CG) based method. Our proposed approximate Hessian matrix is memory-efficient and can be applied to any FCNNs where the activation and criterion functions are twice differentiable. We devise a CG-based method incorporating one-rank approximation to derive Newton directions for training FCNNs, which significantly reduces both space and time complexity. This CG-based method can be employed to solve any linear equation where the coefficient matrix is Kronecker-factored, symmetric and positive definite. Empirical studies show the efficacy and efficiency of our proposed 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。

↑