IMPORTANCE SAMPLING FOR RARE-EVENT GRADIENT ESTIMATION

被引：0

作者：

Bai, Yuanlu ^{[1
]}

He, Shengyi ^{[1
]}

Lam, Henry ^{[1
]}

Jiang, Guangxin ^{[2
]}

Fu, Michael C. ^{[3
,4
]}

机构：

[1] Columbia Univ, Dept Ind Engn & Operat Res, 500 W 120th St, New York, NY 10027 USA

[2] Harbin Inst Technol, Sch Management, 92 Xidazhi St, Harbin 150001, Heilongjiang, Peoples R China

[3] Univ Maryland, Robert H Smith Sch Business, College Pk, MD 20742 USA

[4] Univ Maryland, Inst Syst Res, College Pk, MD 20742 USA

来源：

2022 WINTER SIMULATION CONFERENCE (WSC) | 2022年

基金：

中国国家自然科学基金; 美国国家科学基金会;

关键词：

STOCHASTIC DERIVATIVE ESTIMATOR; SENSITIVITY ANALYSIS; OPTIMIZATION; SIMULATIONS;

D O I：

暂无

中图分类号：

TP301 [理论、方法];

学科分类号：

081202 ;

摘要：

Importance sampling (IS) is a powerful tool for rare-event estimation. However, in many settings, we need to estimate not only the performance expectation but also its gradient. In this paper, we build a bridge from the IS for rare-event estimation to gradient estimation. We establish that, for a class of problems, an efficient IS sampler for estimating the probability of the underlying rare event is also efficient for estimating gradients of expectations over the same rare-event set. We show that both the infinitesimal perturbation analysis and the likelihood ratio estimators can be studied under the proposed framework. We use two numerical examples to validate our findings.

引用

页码：3063 / 3074

页数：12

共 26 条

[1] Bai Y., 2022, ACM T MODELING COMPU
[2] Blanchet J, 2012, SURVEYS OPERAT RES M, V17, P38, DOI [10.1016/j.sorms.2011.09.002, DOI 10.1016/J.S0RMS.2011.09.002]
[3] Bucklew J. A., 2004, INTRO RARE EVENT SIM
[4] Accelerating variance-reduced stochastic gradient methods
Driggs, Derek
Ehrhardt, Matthias J.
Schonlieb, Carola-Bibiane
[J]. MATHEMATICAL PROGRAMMING, 2022, 191 (02) : 671 - 715
[5] Fox L, 1989, PROBAB ENG INFORM SC, V3, P299
[6] Fu M., 1997, CONDITIONAL MONTE CA
[7] Fu MC, 2006, HBK OPERAT RES MANAG, V13, P575, DOI 10.1016/S0927-0507(06)13019-4
[8] Glasserman P., 1988, 1988 Winter Simulation Conference Proceedings (IEEE Cat. No.88CG2659-1), P518, DOI 10.1109/WSC.1988.716212
[9] OPTIMIZATION OF STOCHASTIC-SYSTEMS VIA SIMULATION
GLYNN, PW
[J]. 1989 WINTER SIMULATION CONFERENCE PROCEEDINGS, 1989, : 90 - 105
[10] LIKELIHOOD RATIO GRADIENT ESTIMATION FOR STOCHASTIC-SYSTEMS
GLYNN, PW
[J]. COMMUNICATIONS OF THE ACM, 1990, 33 (10) : 75 - 84

← 1 2 3 →