arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2103.03265cs.LGstat.ML
收藏

使用二阶动量的更好的SGD

Better SGD using Second-order Momentum

Boston University(波士顿大学)

Hoang Tran, Ashok Cutkosky

摘要:

我们开发了一个新的算法,非凸随机优化,找到一个$ε$ -临界点的最佳$O(ε^{-3})$随机梯度和海森向量积计算。我们的算法使用海森矢量积来“校正”SGD动量中的偏置项。这导致更好的梯度估计的方式类似于方差减少方法。与以前的工作相比,我们不需要过大的批量大小,并能够提供一个自适应算法,其收敛速度自动提高,降低梯度估计的方差。我们在各种大规模深度学习架构和基准测试任务上验证了我们的结果。

英文摘要:

We develop a new algorithm for non-convex stochastic optimization that finds an $ε$-critical point in the optimal $O(ε^{-3})$ stochastic gradient and Hessian-vector product computations. Our algorithm uses Hessian-vector products to "correct" a bias term in the momentum of SGD with momentum. This leads to better gradient estimates in a manner analogous to variance reduction methods. In contrast to prior work, we do not require excessively large batch sizes, and are able to provide an adaptive algorithm whose convergence rate automatically improves with decreasing variance in the gradient estimates. We validate our results on a variety of large-scale deep learning architectures and benchmarks tasks.

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。

↑