Tail Uncertainty Analysis in Complex Systems

Abstract The paper presents an efficient computational method for estimating the tails of a target variable Z which is related to other set of bounded variables X = ( X 1 ,…, X n ) by an increasing (decreasing) relation Z = h ( X 1 ,…, X n ). To this aim, variables X i , i = 1,…, n are sequentially simulated in such a manner that Z = h ( x 1 ,…, x i − 1 , X i ,…, X n ) is guaranteed to be in the tail of Z . The method is shown to be very useful to perform an uncertainty analysis of Bayesian networks, when very large confidence intervals for the marginal/conditional probabilities are required, as in reliability or risk analysis. The method is shown to behave best when all scores coincide and is illustrated with several examples, including two examples of application to real cases. A comparison with the fast probability integration method, the best known method to date for solving this problem, shows that it gives better approximations.

[1]  J. Pickands.  Statistical Inference Using Extreme Order Statistics , 1975 .

[2]  F. E. Haskin,et al.  Analysis of extreme top event frequency percentiles based on fast probability integration , 1993 .

[3]  Helmut Prendinger,et al.  Approximate Reasoning , 1997, EPIA.

[4]  R. Rackwitz,et al.  Structural reliability under combined random load sequences , 1978 .

[5]  L. N. Kanal,et al.  Uncertainty in Artificial Intelligence 5 , 1990 .

[6]  I. Weissman.  Estimation of Parameters and Large Quantiles Based on the k Largest Observations , 1978 .

[7]  Judea Pearl,et al.  Evidential Reasoning Using Stochastic Simulation of Causal Models , 1987, Artif. Intell..

[8]  K. Breitung.  Asymptotic approximations for multinormal integrals , 1984 .

[9]  A. M. Hasofer,et al.  Exact and Invariant Second-Moment Code Format , 1974 .

[10]  Ross D. Shachter,et al.  Simulation Approaches to General Probabilistic Inference on Belief Networks , 2013, UAI.

[11]  Enrique F. Castillo,et al.  A modified simulation scheme for inference in Bayesian networks , 1996, Int. J. Approx. Reason..

[12]  Enrique F. Castillo,et al.  Expert Systems and Probabilistic Network Models , 1996, Monographs in Computer Science.

[13]  J. D. T. Oliveira,et al.  The Asymptotic Theory of Extreme Order Statistics , 1979 .

[14]  F. E. Haskin,et al.  Efficient uncertainty analyses using fast probability integration , 1996 .

[15]  Alfred M. Freudenthal,et al.  SAFETY AND THE PROBABILITY OF STRUCTURAL FAILURE , 1956 .

[16]  Enrique F. Castillo,et al.  Parametric Structure of Probabilities in Bayesian Networks , 1995, ECSQARU.

[17]  L. Devroye.  Non-Uniform Random Variate Generation , 1986 .

[18]  Gregory F. Cooper,et al.  A randomized approximation algorithm for probabilistic inference on bayesian belief networks , 1990, Networks.

[19]  Enrique Castillo,et al.  Conditionally Specified Distributions , 1992 .

[20]  Reuven Y. Rubinstein,et al.  Simulation and the Monte Carlo Method , 1981 .

[21]  Enrique Castillo.  Extreme value theory in engineering , 1988 .