Optimal decorrelated score subsampling for generalized linear models with massive data

Optimal decorrelated score subsampling for generalized linear models with massive data
复制标题

海量数据广义线性模型的最优解相关分数子采样

DOI:
10.1007/s11425-022-2057-8
复制
发表时间:
2023-06
期刊:
SCIENCE CHINA Mathematics
影响因子:
--
通讯作者:
Heng Lain
Heng Lain
中科院分区:
其他
文献类型:
--
作者:
Junzhuo Gao;Lei Wang;Heng Lain

文献摘要

参考文献

相似文献

本文研究了具有大量数据的低/高维广义线性模型(GLMs)在存在干扰参数的情况下,对主要感兴趣的低维参数的统一最优子抽样估计和推断。我们首先提出了一个通用的子采样去相关分数函数,以减少不准确的干扰参数估计和缓慢的收敛速度的影响。建立了一般去相关分数子抽样算法所得子样本估计量的一致性和渐近正态性,并在a -和l -最优准则下推导了两个最优子抽样概率,以减小数据量和减少计算量。所提出的最优子抽样概率可证明提高了低维glm中子抽样方案的渐近效率,并优于高维glm中的均匀子抽样方案。进一步提出了一种两步算法,并给出了相应估计量的渐近性质。仿真结果表明,所提估计器的性能令人满意,并在人口普查收入和时尚- mnist数据集上的两个应用也证明了它的实用性。
In this paper, we consider the unified optimal subsampling estimation and inference on the low-dimensional parameter of main interest in the presence of the nuisance parameter for low/high-dimensional generalized linear models (GLMs) with massive data. We first present a general subsampling decorrelated score function to reduce the influence of the less accurate nuisance parameter estimation with the slow convergence rate. The consistency and asymptotic normality of the resultant subsample estimator from a general decorrelated score subsampling algorithm are established, and two optimal subsampling probabilities are derived under the A- and L-optimality criteria to downsize the data volume and reduce the computational burden. The proposed optimal subsampling probabilities provably improve the asymptotic efficiency upon the subsampling schemes in the low-dimensional GLMs and perform better than the uniform subsampling scheme in the high-dimensional GLMs. A two-step algorithm is further proposed to implement and the asymptotic properties of the corresponding estimators are also given. Simulations show satisfactory performance of the proposed estimators, and two applications to census income and Fashion-MNIST datasets also demonstrate its practical applicability.
DOI: 10.1080/00401706.2016.1142900
发表时间: 2016
期刊: Technometrics : a journal of statistics for the physical, chemical, and engineering sciences
影响因子: --
作者:
Schifano ED;Wu J;Wang C;Yan J;Chen MH
通讯作者: Chen MH
DOI: 10.5555/1756006.1859929
发表时间: 2010-03
期刊: J. Mach. Learn. Res.
影响因子: --
作者:
Garvesh Raskutti;M. Wainwright;Bin Yu
通讯作者: Garvesh Raskutti;M. Wainwright;Bin Yu
DOI: --
发表时间: 2020-02
期刊: ArXiv
影响因子: --
作者:
Ping Ma;Xinlian Zhang;Xin Xing;Jingyi Ma;Michael W. Mahoney
通讯作者: Ping Ma;Xinlian Zhang;Xin Xing;Jingyi Ma;Michael W. Mahoney
DOI: 10.5555/2789272.2831141
发表时间: 2013-06
期刊: ArXiv
影响因子: --
作者:
Ping Ma;Michael W. Mahoney;Bin Yu
通讯作者: Ping Ma;Michael W. Mahoney;Bin Yu
DOI: 10.5705/ss.202019.0315
发表时间: 2022-01-01
期刊: STATISTICA SINICA
影响因子: 1.4
作者:
Zhang, Huiming;Jia, Jinzhu
通讯作者: Jia, Jinzhu