arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 1603.00570cs.LGmath.OCstat.ML
收藏

随机梯度法的无替换抽样:收敛结果及其在分布式优化中的应用

Without-Replacement Sampling for Stochastic Gradient Methods: Convergence Results and Application to Distributed Optimization

Weizmann Institute of Science(魏茨曼科学研究学院)

Ohad Shamir

摘要:

机器学习和优化问题的随机梯度方法通常假设数据点被采样\emph{with}替换。但在实践中,采样\emph{without}替换非常常见,在许多情况下更容易实现,并且通常性能更好。在本文中,我们在各种场景下为三种类型的算法提供了无替换抽样的竞争性收敛保证:任何具有在线遗憾保证的算法、随机梯度下降和SVRG。我们的SVRG分析的一个有用的应用是一个接近最优的算法,正则化最小二乘在分布式设置,在通信复杂性和运行时复杂性,当数据被随机分区和条件数可以大到每台机器的数据大小(对数因子)。我们的证明技术结合了随机优化,对抗性在线学习和转导学习理论的思想,并且可以潜在地应用于其他随机优化和学习问题。

英文摘要:

Stochastic gradient methods for machine learning and optimization problems are usually analyzed assuming data points are sampled \emph{with} replacement. In practice, however, sampling \emph{without} replacement is very common, easier to implement in many cases, and often performs better. In this paper, we provide competitive convergence guarantees for without-replacement sampling, under various scenarios, for three types of algorithms: Any algorithm with online regret guarantees, stochastic gradient descent, and SVRG. A useful application of our SVRG analysis is a nearly-optimal algorithm for regularized least squares in a distributed setting, in terms of both communication complexity and runtime complexity, when the data is randomly partitioned and the condition number can be as large as the data size per machine (up to logarithmic factors). Our proof techniques combine ideas from stochastic optimization, adversarial online learning, and transductive learning theory, and can potentially be applied to other stochastic optimization and learning problems.

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。

↑