This paper considers distributed M-estimation under heterogeneous distributions among distributed data blocks. A weighted distributed estimator is proposed to improve the efficiency of the standard "Split-And-Conquer" (SaC) estimator for the common parameter shared by all the data blocks. The weighted distributed estimator is shown to be at least as efficient as the would-be full sample and the generalized method of moment estimators with the latter two estimators requiring full data access. A bias reduction is formulated to the WD estimator to accommodate much larger numbers of data blocks than the existing methods without sacrificing the estimation efficiency, and a similar debiased operation is made to the SaC estimator. The mean squared error (MSE) bounds and the asymptotic distributions of the WD and the two debiased estimators are derived, which shows advantageous performance of the debiased estimators when the number of data blocks is large.
翻译:本文根据分布式数据区块之间分布式分布式分布式分布式分布式估计,提出加权分布式估计,以提高所有数据区块共享的共同参数标准“Split-And-Conquer”(SaC)估计值的效率。加权分布式估计值显示至少与预期的完整抽样和时空估计器的普遍方法一样有效,后者是需要完全数据访问的两个估计值。向WD估计值设定了偏差,以便在不牺牲估计效率的情况下容纳比现有方法多得多的数据区块,并对SaC估计值做了类似的偏差操作。平均正方形误差(MSE)界限和WD和两个偏差估计值偏差估计值的偏差分布是推断出来的,这表明在数据区块数量较大时,偏差估计值的估算值的偏差性表现是有利的。