arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2011.02761cs.DScs.ITcs.LGmath.ITstat.COstat.ML
收藏

基于实例的剖面最大似然估计

Instance Based Approximations to Profile Maximum Likelihood

Stanford University(斯坦福大学)

Nima Anari, Moses Charikar, Kirankumar Shiragur, Aaron Sidford

摘要:

在本文中,我们提供了一个新的有效算法近似计算轮廓最大似然(PML)分布,对称属性估计中的一个突出的数量。我们提供了一个算法,它匹配以前最好的已知的高效算法计算近似PML分布和改进时,在给定的情况下,不同的观察到的频率的数量是小的。我们实现这一结果,利用新的稀疏结构的近似PML分布,并提供了一个新的矩阵舍入算法,独立的利益。利用这一结果,我们获得了第一个可证明的计算效率的实现伪PML,一个通用的框架,估计广泛的对称属性。此外,我们获得了有效的基于PML的估计分布与小轮廓熵,一个自然的基于实例的复杂性措施。此外,我们提供了一个更简单,更实用的伪PML实现,匹配最知名的理论保证这样的估计和评估这种方法的经验。

英文摘要:

In this paper we provide a new efficient algorithm for approximately computing the profile maximum likelihood (PML) distribution, a prominent quantity in symmetric property estimation. We provide an algorithm which matches the previous best known efficient algorithms for computing approximate PML distributions and improves when the number of distinct observed frequencies in the given instance is small. We achieve this result by exploiting new sparsity structure in approximate PML distributions and providing a new matrix rounding algorithm, of independent interest. Leveraging this result, we obtain the first provable computationally efficient implementation of PseudoPML, a general framework for estimating a broad class of symmetric properties. Additionally, we obtain efficient PML-based estimators for distributions with small profile entropy, a natural instance-based complexity measure. Further, we provide a simpler and more practical PseudoPML implementation that matches the best-known theoretical guarantees of such an estimator and evaluate this method empirically.

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。

↑