In the past few years, approximate Bayesian Neural Networks (BNNs) have demonstrated the ability to produce statistically consistent posteriors on a wide range of inference problems at unprecedented speed and scale. However, any disconnect between training sets and the distribution of real-world objects can introduce bias when BNNs are applied to data. This is a common challenge in astrophysics and cosmology, where the unknown distribution of objects in our Universe is often the science goal. In this work, we incorporate BNNs with flexible posterior parameterizations into a hierarchical inference framework that allows for the reconstruction of population hyperparameters and removes the bias introduced by the training distribution. We focus on the challenge of producing posterior PDFs for strong gravitational lens mass model parameters given Hubble Space Telescope (HST) quality single-filter, lens-subtracted, synthetic imaging data. We show that the posterior PDFs are sufficiently accurate (i.e., statistically consistent with the truth) across a wide variety of power-law elliptical lens mass distributions. We then apply our approach to test data sets whose lens parameters are drawn from distributions that are drastically different from the training set. We show that our hierarchical inference framework mitigates the bias introduced by an unrepresentative training set's interim prior. Simultaneously, given a sufficiently broad training set, we can precisely reconstruct the population hyperparameters governing our test distributions. Our full pipeline, from training to hierarchical inference on thousands of lenses, can be run in a day. The framework presented here will allow us to efficiently exploit the full constraining power of future ground- and space-based surveys.
[1]
M. Barnabe,et al.
TDCOSMO IV: Hierarchical time-delay cosmography -- joint inference of the Hubble constant and galaxy density profiles
,
2020
.
[2]
A. R. Rao.
Annual Review of Astronomy and Astrophysics
,
2015
.
[3]
Chaoping Xing,et al.
New MDS Self-Dual Codes From Generalized Reed—Solomon Codes
,
2016,
IEEE Transactions on Information Theory.
[4]
S. Suyu,et al.
HOLISMOKES -- IV. Efficient Mass Modeling of Strong Lenses through Deep Learning.
,
2020,
2010.00602.
[5]
David S. Greenberg,et al.
Automatic Posterior Transformation for Likelihood-Free Inference
,
2019,
ICML.
[6]
Ralph C. Bohlin,et al.
New Estimates of the Sky Background for the HST Exposure Time Calculator
,
2002
.
[7]
Zoubin Ghahramani,et al.
Deep Bayesian Active Learning with Image Data
,
2017,
ICML.
[8]
Dandan Xu,et al.
Time Delay Lens Modeling Challenge: I. Experimental Design
,
2018,
1801.01506.
[9]
T. Treu,et al.
TDCOSMO V: strategies for precise and accurate measurements of the Hubble constant with strong lensing
,
2020,
2008.06157.
[10]
Laurence Perreault Levasseur,et al.
Bayesian Neural Networks
,
2020,
Artificial Intelligence for High Energy Physics.
[11]
Zoubin Ghahramani,et al.
Bayesian Convolutional Neural Networks with Bernoulli Approximate Variational Inference
,
2015,
ArXiv.
[12]
Ana Diaz Rivero,et al.
Detecting Subhalos in Strong Gravitational Lens Images with Image Segmentation
,
2020
.