Please use this identifier to cite or link to this item: https://doi.org/10.21256/zhaw-25627
Publication type: Article in scientific journal
Type of review: Peer review (publication)
Title: Learning summary statistics for Bayesian inference with autoencoders
Authors: Albert, Carlo
Ulzega, Simone
Ozdemir, Firat
Perez-Cruz, Fernando
Mira, Antonietta
et. al: No
DOI: 10.21468/SciPostPhysCore.5.3.043
10.21256/zhaw-25627
Published in: SciPost Physics Core
Volume(Issue): 5
Issue: 3
Page(s): 043
Issue Date: 2022
Publisher / Ed. Institution: SciPost Foundation
ISSN: 2666-9366
Language: English
Subjects: Bayesian inference; Machine learning; Deep neural network; Autoencoder; Dimensionality reduction
Subject (DDC): 006: Special computer methods
510: Mathematics
Abstract: For stochastic models with intractable likelihood functions, approximate Bayesian computation offers a way of approximating the true posterior through repeated comparisons of observations with simulated model outputs in terms of a small set of summary statistics. These statistics need to retain the information that is relevant for constraining the parameters but cancel out the noise. They can thus be seen as thermodynamic state variables, for general stochastic models. For many scientific applications, we need strictly more summary statistics than model parameters to reach a satisfactory approximation of the posterior. Therefore, we propose to use a latent representation of deep neural networks based on Autoencoders as summary statistics. To create an incentive for the encoder to encode all the parameter-related information but not the noise, we give the decoder access to explicit or implicit information on the noise that has been used to generate the training data. We validate the approach empirically on two types of stochastic models.
URI: https://digitalcollection.zhaw.ch/handle/11475/25627
Fulltext version: Published version
License (according to publishing contract): CC BY 4.0: Attribution 4.0 International
Departement: Life Sciences and Facility Management
Organisational Unit: Institute of Computational Life Sciences (ICLS)
Published as part of the ZHAW project: BISTOM - Bayesian Inference with Stochastic Models
Appears in collections:Publikationen Life Sciences und Facility Management

Files in This Item:
File Description SizeFormat 
2022_Albert-etal_Learning-summary-statistics-Bayesian-inference-autoencoders.pdf4.57 MBAdobe PDFThumbnail
View/Open
Show full item record
Albert, C., Ulzega, S., Ozdemir, F., Perez-Cruz, F., & Mira, A. (2022). Learning summary statistics for Bayesian inference with autoencoders. SciPost Physics Core, 5(3), 43. https://doi.org/10.21468/SciPostPhysCore.5.3.043
Albert, C. et al. (2022) ‘Learning summary statistics for Bayesian inference with autoencoders’, SciPost Physics Core, 5(3), p. 043. Available at: https://doi.org/10.21468/SciPostPhysCore.5.3.043.
C. Albert, S. Ulzega, F. Ozdemir, F. Perez-Cruz, and A. Mira, “Learning summary statistics for Bayesian inference with autoencoders,” SciPost Physics Core, vol. 5, no. 3, p. 043, 2022, doi: 10.21468/SciPostPhysCore.5.3.043.
ALBERT, Carlo, Simone ULZEGA, Firat OZDEMIR, Fernando PEREZ-CRUZ und Antonietta MIRA, 2022. Learning summary statistics for Bayesian inference with autoencoders. SciPost Physics Core. 2022. Bd. 5, Nr. 3, S. 043. DOI 10.21468/SciPostPhysCore.5.3.043
Albert, Carlo, Simone Ulzega, Firat Ozdemir, Fernando Perez-Cruz, and Antonietta Mira. 2022. “Learning Summary Statistics for Bayesian Inference with Autoencoders.” SciPost Physics Core 5 (3): 43. https://doi.org/10.21468/SciPostPhysCore.5.3.043.
Albert, Carlo, et al. “Learning Summary Statistics for Bayesian Inference with Autoencoders.” SciPost Physics Core, vol. 5, no. 3, 2022, p. 43, https://doi.org/10.21468/SciPostPhysCore.5.3.043.


Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.