Semiparametric Cross Entropy for Rare-Event Simulation

The Cross Entropy method is a well-known adaptive importance sampling method for rare-event probability estimation, which requires estimating an optimal importance sampling density within a parametric class. In this article we estimate an optimal importance sampling density within a wider semiparametric class of distributions. We show that this semiparametric version of the Cross Entropy method frequently yields efficient estimators. We illustrate the excellent practical performance of the method with numerical experiments and show that for the problems we consider it typically outperforms alternative schemes by orders of magnitude.

[1]  Jürgen Hartinger,et al.  On the efficiency of the Asmussen–Kroese-estimator and its application to stop-loss transforms , 2009 .

[2]  Improving the Asmussen-Kroese Type Simulation Estimators , 2012 .

[3]  Peter W. Glynn,et al.  A comparison of cross-entropy and variance minimization strategies , 2011, Journal of Applied Probability.

[4]  P. Embrechts,et al.  On closure and factorization properties of subexponential and related distributions , 1980, Journal of the Australian Mathematical Society. Series A. Pure Mathematics and Statistics.

[5]  Dirk P. Kroese,et al.  Efficient Monte Carlo simulation via the generalized splitting method , 2012, Stat. Comput..

[6]  Søren Asmussen,et al.  On error rates in rare event simulation with heavy tails , 2012, Proceedings Title: Proceedings of the 2012 Winter Simulation Conference (WSC).

[7]  Sandeep Juneja,et al.  Estimating tail probabilities of heavy tailed distributions with asymptotically zero relative error , 2007, Queueing Syst. Theory Appl..

[8]  Dirk P. Kroese,et al.  Handbook of Monte Carlo Methods , 2011 .

[9]  Dirk P. Kroese,et al.  The cross-entropy method for estimation , 2013 .

[10]  Bruno Tuffin,et al.  Probabilistic bounded relative error for rare event simulation learning techniques , 2012, Proceedings Title: Proceedings of the 2012 Winter Simulation Conference (WSC).

[11]  Ioannis Ntzoufras,et al.  On the use of marginal posteriors in marginal likelihood estimation via importance sampling , 2013, Comput. Stat. Data Anal..

[12]  Bruno Tuffin,et al.  Markov chain importance sampling with applications to rare event probability estimation , 2011, Stat. Comput..

[13]  Peter W. Glynn,et al.  Asymptotic robustness of estimators in rare-event simulation , 2007, TOMC.

[14]  Paul Dupuis,et al.  Importance sampling for sums of random variables with regularly varying tails , 2007, TOMC.

[15]  Dirk P. Kroese,et al.  The Cross-Entropy Method: A Unified Approach to Combinatorial Optimization, Monte-Carlo Simulation and Machine Learning , 2004 .

[16]  Dirk P. Kroese,et al.  HEAVY TAILS, IMPORTANCE SAMPLING AND CROSS–ENTROPY , 2005 .

[17]  Dirk P. Kroese,et al.  Improved cross-entropy method for estimation , 2011, Statistics and Computing.

[18]  C. Klüppelberg,et al.  Modelling Extremal Events , 1997 .

[19]  Peter W. Glynn,et al.  Stochastic Simulation: Algorithms and Analysis , 2007 .

[20]  Dirk P. Kroese,et al.  Improved algorithms for rare event simulation with heavy tails , 2006, Advances in Applied Probability.

[21]  Dirk P. Kroese,et al.  The Cross Entropy Method: A Unified Approach To Combinatorial Optimization, Monte-carlo Simulation (Information Science and Statistics) , 2004 .

[22]  S. Foss,et al.  An Introduction to Heavy-Tailed and Subexponential Distributions , 2011 .