High-Dimensional Robust Mean Estimation via Gradient Descent

High-Dimensional Robust Mean Estimation via Gradient Descent
复制标题

DOI:
--
复制
发表时间:
2020-05
期刊:
--
影响因子:
--
通讯作者:
Yu Cheng;Ilias Diakonikolas;Rong Ge;M. Soltanolkotabi
Yu Cheng;Ilias Diakonikolas;Rong Ge;M. Soltanolkotabi
中科院分区:
其他
文献类型:
--
作者:
Yu Cheng;Ilias Diakonikolas;Rong Ge;M. Soltanolkotabi

文献摘要

被引文献

相似文献

我们研究了高维鲁棒均值估计的问题,在存在一个常数分数的敌对离群值。最近的一系列工作提供了复杂的多项式时间算法,这个问题与尺寸无关的误差保证范围内的自然分布家庭。在这项工作中,我们表明,一个自然的非凸制定的问题可以直接解决梯度下降。我们的方法利用了一个新的结构引理,粗略地表明,我们的非凸目标的任何近似稳定点给出了一个接近最优的解决方案,以基本的鲁棒估计任务。我们的工作建立了一个有趣的算法高维鲁棒统计和非凸优化,这可能有更广泛的应用到其他鲁棒估计任务之间的联系。
We study the problem of high-dimensional robust mean estimation in the presence of a constant fraction of adversarial outliers. A recent line of work has provided sophisticated polynomial-time algorithms for this problem with dimension-independent error guarantees for a range of natural distribution families. In this work, we show that a natural non-convex formulation of the problem can be solved directly by gradient descent. Our approach leverages a novel structural lemma, roughly showing that any approximate stationary point of our non-convex objective gives a near-optimal solution to the underlying robust estimation task. Our work establishes an intriguing connection between algorithmic high-dimensional robust statistics and non-convex optimization, which may have broader applications to other robust estimation tasks.