Details
| Original language | English |
|---|---|
| Title of host publication | Proceedings of the 8th International Symposium on Reliability Engineering and Risk Management, ISRERM 2022 |
| Editors | Michael Beer, Enrico Zio, Kok-Kwang Phoon, Bilal M. Ayyub |
| Pages | 173-178 |
| Number of pages | 6 |
| Publication status | Published - 2022 |
| Event | 8th International Symposium on Reliability Engineering and Risk Management, ISRERM 2022 - Hannover, Germany Duration: 4 Sept 2022 → 7 Sept 2022 |
Abstract
In this work we quantify the uncertainty over Power Spectral Density estimation of stochastic processes based on realizations with gapped missing data. For the purpose of imputation, a fully-connected neural network architecture that works in an autoregressive manner is firstly constructed to probabilistically capture the temporal patterns in the time series data. Particularly, under the Bayesian scheme, uncertainties with respect to the parameters of the neural network model (i.e. weights) are introduced by multivariate Gaussian distribution. During training, the posteriors are learnt through variational inference approach. As a result, the missing gaps can be recursively imputed via our neural network in each realization, and thanks to the probabilistic merit of Bayesian inference, an ensemble of reconstructed realizations can then be obtained. Further, by resorting to a Fourier-based spectral estimation method, a probabilistic power spectrum could be derived, with each frequency component represented by a probability distribution.
Keywords
- Bayesian neural network, missing data, spectral estimation, stochastic process, Variational Bayesian inference
ASJC Scopus subject areas
- Decision Sciences(all)
- Management Science and Operations Research
- Engineering(all)
- Safety, Risk, Reliability and Quality
Cite this
- Standard
- Harvard
- Apa
- Vancouver
- BibTeX
- RIS
Proceedings of the 8th International Symposium on Reliability Engineering and Risk Management, ISRERM 2022. ed. / Michael Beer; Enrico Zio; Kok-Kwang Phoon; Bilal M. Ayyub. 2022. p. 173-178.
Research output: Chapter in book/report/conference proceeding › Conference contribution › Research › peer review
}
TY - GEN
T1 - Uncertainty Quantification Over Spectral Estimation of Stochastic Processes Subject to Gapped Missing Data Using Variational Bayesian Inference
AU - Chen, Yu
AU - Patelli, Edoardo
AU - Beer, Michael
AU - Edwards, Ben
N1 - Publisher Copyright: © 2022 ISRERM Organizers. Published by Research Publishing, Singapore.
PY - 2022
Y1 - 2022
N2 - In this work we quantify the uncertainty over Power Spectral Density estimation of stochastic processes based on realizations with gapped missing data. For the purpose of imputation, a fully-connected neural network architecture that works in an autoregressive manner is firstly constructed to probabilistically capture the temporal patterns in the time series data. Particularly, under the Bayesian scheme, uncertainties with respect to the parameters of the neural network model (i.e. weights) are introduced by multivariate Gaussian distribution. During training, the posteriors are learnt through variational inference approach. As a result, the missing gaps can be recursively imputed via our neural network in each realization, and thanks to the probabilistic merit of Bayesian inference, an ensemble of reconstructed realizations can then be obtained. Further, by resorting to a Fourier-based spectral estimation method, a probabilistic power spectrum could be derived, with each frequency component represented by a probability distribution.
AB - In this work we quantify the uncertainty over Power Spectral Density estimation of stochastic processes based on realizations with gapped missing data. For the purpose of imputation, a fully-connected neural network architecture that works in an autoregressive manner is firstly constructed to probabilistically capture the temporal patterns in the time series data. Particularly, under the Bayesian scheme, uncertainties with respect to the parameters of the neural network model (i.e. weights) are introduced by multivariate Gaussian distribution. During training, the posteriors are learnt through variational inference approach. As a result, the missing gaps can be recursively imputed via our neural network in each realization, and thanks to the probabilistic merit of Bayesian inference, an ensemble of reconstructed realizations can then be obtained. Further, by resorting to a Fourier-based spectral estimation method, a probabilistic power spectrum could be derived, with each frequency component represented by a probability distribution.
KW - Bayesian neural network
KW - missing data
KW - spectral estimation
KW - stochastic process
KW - Variational Bayesian inference
UR - http://www.scopus.com/inward/record.url?scp=85202046515&partnerID=8YFLogxK
U2 - 10.3850/978-981-18-5184-1_MS-06-179-cd
DO - 10.3850/978-981-18-5184-1_MS-06-179-cd
M3 - Conference contribution
AN - SCOPUS:85202046515
SN - 9789811851841
SP - 173
EP - 178
BT - Proceedings of the 8th International Symposium on Reliability Engineering and Risk Management, ISRERM 2022
A2 - Beer, Michael
A2 - Zio, Enrico
A2 - Phoon, Kok-Kwang
A2 - Ayyub, Bilal M.
T2 - 8th International Symposium on Reliability Engineering and Risk Management, ISRERM 2022
Y2 - 4 September 2022 through 7 September 2022
ER -