TY  - JOUR
PB  - SciPost Foundation
DO  - 10.21468/SciPostPhysCore.5.3.043
TI  - Learning Summary Statistics for Bayesian Inference with Autoencoders
PY  - 2022/09/01
UR  - https://scipost.org/SciPostPhysCore.5.3.043
JF  - SciPost Physics Core
JA  - SciPost Phys. Core
VL  - 5
IS  - 3
SP  - 043
A1  - Albert, Carlo
AU  - Ulzega, Simone
AU  - Ozdemir, Firat
AU  - Perez-Cruz, Fernando
AU  - Mira, Antonietta
AB  - 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 the inner dimension of deep neural network based 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.
ER  -

@Article{10.21468/SciPostPhysCore.5.3.043,
	title={{Learning Summary Statistics for Bayesian Inference with Autoencoders}},
	author={Carlo Albert and Simone Ulzega and Firat Ozdemir and Fernando Perez-Cruz and Antonietta Mira},
	journal={SciPost Phys. Core},
	volume={5},
	pages={043},
	year={2022},
	publisher={SciPost},
	doi={10.21468/SciPostPhysCore.5.3.043},
	url={https://scipost.org/10.21468/SciPostPhysCore.5.3.043},
}