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 1709.05069cs.LG
收藏

使用近似Hessian矩阵加速用于分布式深度学习的SGD

Accelerating SGD for Distributed Deep-Learning Using Approximated Hessian Matrix

University of Southern California(南加州大学)

Sébastien M. R. Arnold, Chunming Wang

摘要:

我们介绍了一种新的方法来计算一个秩$m$近似的海森矩阵的逆在分布式政权。通过利用多个Workers的梯度和参数的差异,我们能够有效地实现Newton-Raphson方法的分布式近似。我们还提出了初步的结果,强调大型随机优化问题的二阶方法的优势和挑战。特别是,我们的工作表明,结合梯度的新策略提供了损失表面的进一步信息。

英文摘要:

We introduce a novel method to compute a rank $m$ approximation of the inverse of the Hessian matrix in the distributed regime. By leveraging the differences in gradients and parameters of multiple Workers, we are able to efficiently implement a distributed approximation of the Newton-Raphson method. We also present preliminary results which underline advantages and challenges of second-order methods for large stochastic optimization problems. In particular, our work suggests that novel strategies for combining gradients provide further information on the loss surface.

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。

↑