Cross entropy-based importance sampling using Gaussian densities revisited

•We compare two distributions as sampling densities in the cross entropy method.•A modified expectation-maximization algorithm is proposed for fitting the parameters.•The DBSCAN algorithm is used to estimate the number of Gaussians in the mixture.•We compare the performance of the two densities in t...

Celý popis

Uloženo v:

Podrobná bibliografie
Vydáno v:	Structural safety Ročník 76; s. 15 - 27
Hlavní autoři:	Geyer, Sebastian, Papaioannou, Iason, Straub, Daniel
Médium:	Journal Article
Jazyk:	angličtina
Vydáno:	Amsterdam Elsevier Ltd 01.01.2019 Elsevier BV
Témata:	Adaptive sampling Algorithms Clustering Component reliability Computer simulation Cross entropy method Density Divergence Entropy Entropy (Information theory) Expectation maximization Failure analysis Gaussian mixture Importance sampling Markov chains Mathematical models Normal distribution Optimization Probabilistic models Probability Reliability analysis Reliability engineering Sampling Sequential importance sampling Simulation Statistical analysis Structural reliability System reliability Sequential importance sampling Simulation Cross entropy method Expectation maximization Reliability analysis Gaussian mixture Importance sampling
ISSN:	0167-4730, 1879-3355
On-line přístup:	Získat plný text
Tagy:	Přidat tag Žádné tagy, Buďte první, kdo vytvoří štítek k tomuto záznamu!

Buďte první, kdo okomentuje tento záznam!