Distortion estimates for approximate Bayesian inference

Hanwen Xing, Geoff Nicholls, Jeong (Kate) Lee
Proceedings of the 36th Conference on Uncertainty in Artificial Intelligence (UAI), PMLR 124:1208-1217, 2020.

Abstract

Current literature on posterior approximation for Bayesian inference offers many alternative methods. Does our chosen approximation scheme work well on the observed data? The best existing generic diagnostic tools treating this kind of question by looking at performance averaged over data space, or otherwise lack diagnostic detail. However, if the approximation is bad for most data, but good at the observed data, then we may discard a useful approximation. We give graphical diagnostics for posterior approximation at the observed data. We estimate a “distortion map” that acts on univariate marginals of the approximate posterior to move them closer to the exact posterior, without recourse to the exact posterior.

Cite this Paper


BibTeX
@InProceedings{pmlr-v124-xing20b, title = {Distortion estimates for approximate Bayesian inference}, author = {Xing, Hanwen and Nicholls, Geoff and (Kate) Lee, Jeong}, booktitle = {Proceedings of the 36th Conference on Uncertainty in Artificial Intelligence (UAI)}, pages = {1208--1217}, year = {2020}, editor = {Jonas Peters and David Sontag}, volume = {124}, series = {Proceedings of Machine Learning Research}, month = {03--06 Aug}, publisher = {PMLR}, pdf = {http://proceedings.mlr.press/v124/xing20b/xing20b.pdf}, url = { http://proceedings.mlr.press/v124/xing20b.html }, abstract = {Current literature on posterior approximation for Bayesian inference offers many alternative methods. Does our chosen approximation scheme work well on the observed data? The best existing generic diagnostic tools treating this kind of question by looking at performance averaged over data space, or otherwise lack diagnostic detail. However, if the approximation is bad for most data, but good at the observed data, then we may discard a useful approximation. We give graphical diagnostics for posterior approximation at the observed data. We estimate a “distortion map” that acts on univariate marginals of the approximate posterior to move them closer to the exact posterior, without recourse to the exact posterior.} }
Endnote
%0 Conference Paper %T Distortion estimates for approximate Bayesian inference %A Hanwen Xing %A Geoff Nicholls %A Jeong (Kate) Lee %B Proceedings of the 36th Conference on Uncertainty in Artificial Intelligence (UAI) %C Proceedings of Machine Learning Research %D 2020 %E Jonas Peters %E David Sontag %F pmlr-v124-xing20b %I PMLR %P 1208--1217 %U http://proceedings.mlr.press/v124/xing20b.html %V 124 %X Current literature on posterior approximation for Bayesian inference offers many alternative methods. Does our chosen approximation scheme work well on the observed data? The best existing generic diagnostic tools treating this kind of question by looking at performance averaged over data space, or otherwise lack diagnostic detail. However, if the approximation is bad for most data, but good at the observed data, then we may discard a useful approximation. We give graphical diagnostics for posterior approximation at the observed data. We estimate a “distortion map” that acts on univariate marginals of the approximate posterior to move them closer to the exact posterior, without recourse to the exact posterior.
APA
Xing, H., Nicholls, G. & (Kate) Lee, J.. (2020). Distortion estimates for approximate Bayesian inference. Proceedings of the 36th Conference on Uncertainty in Artificial Intelligence (UAI), in Proceedings of Machine Learning Research 124:1208-1217 Available from http://proceedings.mlr.press/v124/xing20b.html .

Related Material