Automatic Posterior Transformation for Likelihood-Free Inference

David Greenberg, Marcel Nonnenmacher, Jakob Macke
Proceedings of the 36th International Conference on Machine Learning, PMLR 97:2404-2414, 2019.

Abstract

How can one perform Bayesian inference on stochastic simulators with intractable likelihoods? A recent approach is to learn the posterior from adaptively proposed simulations using neural network-based conditional density estimators. However, existing methods are limited to a narrow range of proposal distributions or require importance weighting that can limit performance in practice. Here we present automatic posterior transformation (APT), a new sequential neural posterior estimation method for simulation-based inference. APT can modify the posterior estimate using arbitrary, dynamically updated proposals, and is compatible with powerful flow-based density estimators. It is more flexible, scalable and efficient than previous simulation-based inference techniques. APT can operate directly on high-dimensional time series and image data, opening up new applications for likelihood-free inference.

Cite this Paper


BibTeX
@InProceedings{pmlr-v97-greenberg19a, title = {Automatic Posterior Transformation for Likelihood-Free Inference}, author = {Greenberg, David and Nonnenmacher, Marcel and Macke, Jakob}, booktitle = {Proceedings of the 36th International Conference on Machine Learning}, pages = {2404--2414}, year = {2019}, editor = {Chaudhuri, Kamalika and Salakhutdinov, Ruslan}, volume = {97}, series = {Proceedings of Machine Learning Research}, month = {09--15 Jun}, publisher = {PMLR}, pdf = {http://proceedings.mlr.press/v97/greenberg19a/greenberg19a.pdf}, url = {http://proceedings.mlr.press/v97/greenberg19a.html}, abstract = {How can one perform Bayesian inference on stochastic simulators with intractable likelihoods? A recent approach is to learn the posterior from adaptively proposed simulations using neural network-based conditional density estimators. However, existing methods are limited to a narrow range of proposal distributions or require importance weighting that can limit performance in practice. Here we present automatic posterior transformation (APT), a new sequential neural posterior estimation method for simulation-based inference. APT can modify the posterior estimate using arbitrary, dynamically updated proposals, and is compatible with powerful flow-based density estimators. It is more flexible, scalable and efficient than previous simulation-based inference techniques. APT can operate directly on high-dimensional time series and image data, opening up new applications for likelihood-free inference.} }
Endnote
%0 Conference Paper %T Automatic Posterior Transformation for Likelihood-Free Inference %A David Greenberg %A Marcel Nonnenmacher %A Jakob Macke %B Proceedings of the 36th International Conference on Machine Learning %C Proceedings of Machine Learning Research %D 2019 %E Kamalika Chaudhuri %E Ruslan Salakhutdinov %F pmlr-v97-greenberg19a %I PMLR %P 2404--2414 %U http://proceedings.mlr.press/v97/greenberg19a.html %V 97 %X How can one perform Bayesian inference on stochastic simulators with intractable likelihoods? A recent approach is to learn the posterior from adaptively proposed simulations using neural network-based conditional density estimators. However, existing methods are limited to a narrow range of proposal distributions or require importance weighting that can limit performance in practice. Here we present automatic posterior transformation (APT), a new sequential neural posterior estimation method for simulation-based inference. APT can modify the posterior estimate using arbitrary, dynamically updated proposals, and is compatible with powerful flow-based density estimators. It is more flexible, scalable and efficient than previous simulation-based inference techniques. APT can operate directly on high-dimensional time series and image data, opening up new applications for likelihood-free inference.
APA
Greenberg, D., Nonnenmacher, M. & Macke, J.. (2019). Automatic Posterior Transformation for Likelihood-Free Inference. Proceedings of the 36th International Conference on Machine Learning, in Proceedings of Machine Learning Research 97:2404-2414 Available from http://proceedings.mlr.press/v97/greenberg19a.html.

Related Material