Massive parallelization of serial inference algorithms for a complex generalized linear model.

Massive parallelization of serial inference algorithms for a complex generalized linear model.
复制标题

DOI:
10.1145/2414416.2414791
复制
发表时间:
2013-01
期刊:
ACM transactions on modeling and computer simulation : a publication of the Association for Computing Machinery
影响因子:
--
通讯作者:
Madigan D
Madigan D
中科院分区:
其他
文献类型:
--
作者:
Suchard MA;Simpson SE;Zorych I;Ryan P;Madigan D

文献摘要

参考文献

被引文献

相似文献

近年来,在发生了一系列备受瞩目的药品安全灾难后,许多国家都在加倍努力,确保特许医疗产品的安全。在这方面,索赔数据库或电子健康记录系统等大型观测数据库引起了特别注意,但在方法和计算方面存在重大问题。在这篇文章中,我们展示了高性能统计计算,包括图形处理单元,相对廉价的高度并行计算设备,如何使大型数据库中的复杂方法成为可能。我们专注于循环坐标下降方法的优化和大规模并行化,以适应贝叶斯背景下涉及数千万个观测和数千个预报器的条件广义线性模型。我们发现总体运行时间有了数量级的改进。坐标下降方法在高维统计学中无处不在,我们提出的算法开辟了令人兴奋的新方法可能性,有可能显着提高药物安全性。
Following a series of high-profile drug safety disasters in recent years, many countries are redoubling their efforts to ensure the safety of licensed medical products. Large-scale observational databases such as claims databases or electronic health record systems are attracting particular attention in this regard, but present significant methodological and computational concerns. In this paper we show how high-performance statistical computation, including graphics processing units, relatively inexpensive highly parallel computing devices, can enable complex methods in large databases. We focus on optimization and massive parallelization of cyclic coordinate descent approaches to fit a conditioned generalized linear model involving tens of millions of observations and thousands of predictors in a Bayesian context. We find orders-of-magnitude improvement in overall run-time. Coordinate descent approaches are ubiquitous in high-dimensional statistics and the algorithms we propose open up exciting new methodological possibilities with the potential to significantly improve drug safety.
DOI: 10.7326/0003-4819-153-9-201011020-00010
发表时间: 2010-11-02
影响因子: 39.2
作者:
Stang, Paul E.;Ryan, Patrick B.;Woodcock, Janet
通讯作者: Woodcock, Janet
DOI: 10.1109/titb.2007.900808
发表时间: 2008-07-01
影响因子: --
作者:
Jin, Huidong (Warren);Chen, Jie;O'Keefe, Christine M.
通讯作者: O'Keefe, Christine M.
DOI: 10.1097/mlr.0b013e318179253b
发表时间: 2008-09-01
期刊: MEDICAL CARE
影响因子: 3
作者:
Curtis, Jeffrey R.;Cheng, Hong;DuMouchel, William
通讯作者: DuMouchel, William
DOI: 10.1097/ede.0b013e3181a663cc
发表时间: 2009-07
期刊: Epidemiology (Cambridge, Mass.)
影响因子: --
作者:
Schneeweiss S;Rassen JA;Glynn RJ;Avorn J;Mogun H;Brookhart MA
通讯作者: Brookhart MA
DOI: 10.1080/07474946.2011.539924
发表时间: 2011-01-01
影响因子: 0.8
作者:
Kulldorff, Martin;Davis, Robert L.;Platt, Richard
通讯作者: Platt, Richard