Abstract:The estimation of rare event or failure probabilities in high dimensions is of interest in many areas of science and technology. We consider problems where the rare event is expressed in terms of a computationally costly numerical model. Importance sampling with the cross-entropy method offers an efficient way to address such problems provided that a suitable parametric family of biasing densities is employed. Although some existing parametric distribution families are designed to perform efficiently in high dimensions, their applicability within the cross-entropy method is limited to problems with dimension of O(1e2). In this work, rather than directly building sampling densities in high dimensions, we focus on identifying the intrinsic low-dimensional structure of the rare event simulation problem. To this end, we exploit a connection between rare event simulation and Bayesian inverse problems. This allows us to adapt dimension reduction techniques from Bayesian inference to construct new, effectively low-dimensional, biasing distributions within the cross-entropy method. In particular, we employ the approach in [47], as it enables control of the error in the approximation of the optimal biasing distribution. We illustrate our method using two standard high-dimensional reliability benchmark problems and one structural mechanics application involving random fields.

Importance Splitting for Finite-Time Rare Event Simulation

Importance Sampling of Interval Markov Chains

Research of Optimal Splitting Level of Splitting Algorithm

Efficient Importance Sampling for Rare Event Simulation with Applications

Adaptive Reduced Multilevel Splitting

A splitting method to reduce MCMC variance

A self-similarity principle for the computation of rare event probability

Efficient Rare-Event Simulation for Multiple Jump Events in Regularly Varying Lévy Processes with Infinite Activities

A Multi-Level Splitting Algorithm Based on Differential Evolution

A New Method for the Fast Simulation of Models of Highly Dependable Markov System

Propagation of Input Tail Uncertainty in Rare-Event Estimation: A Light versus Heavy Tail Dichotomy

Rare-Event Simulation for Neural Network and Random Forest Predictors

Improving Importance Sampling Method in Structural Reliability

An Importance Sampling Method for Expectation of Portfolio Credit Risk

Metamodel-based importance sampling for the simulation of rare events

Strongly Efficient Rare-Event Simulation for Regularly Varying Lévy Processes with Infinite Activities

Certifiable Deep Importance Sampling for Rare-Event Simulation of Black-Box Systems

Importance Sampling for rare events and conditioned random walks

A unified perspective on exponential tilt and bridge algorithms for rare trajectories of discrete Markov processes

Cross-entropy-based importance sampling with failure-informed dimension reduction for rare event simulation

A Deep Reinforcement Learning Approach to Rare Event Estimation