Celeste: Variational inference for a generative model of astronomical images

Jeffrey Regier, Andrew Miller, Jon McAuliffe, Ryan Adams, Matt Hoffman, Dustin Lang, David Schlegel, Mr Prabhat
Proceedings of the 32nd International Conference on Machine Learning, PMLR 37:2095-2103, 2015.

Abstract

We present a new, fully generative model of optical telescope image sets, along with a variational procedure for inference. Each pixel intensity is treated as a Poisson random variable, with a rate parameter dependent on latent properties of stars and galaxies. Key latent properties are themselves random, with scientific prior distributions constructed from large ancillary data sets. We check our approach on synthetic images. We also run it on images from a major sky survey, where it exceeds the performance of the current state-of-the-art method for locating celestial bodies and measuring their colors.

Cite this Paper

BibTeX
@InProceedings{pmlr-v37-regier15, title = {Celeste: Variational inference for a generative model of astronomical images}, author = {Regier, Jeffrey and Miller, Andrew and McAuliffe, Jon and Adams, Ryan and Hoffman, Matt and Lang, Dustin and Schlegel, David and Prabhat, Mr}, booktitle = {Proceedings of the 32nd International Conference on Machine Learning}, pages = {2095--2103}, year = {2015}, editor = {Bach, Francis and Blei, David}, volume = {37}, series = {Proceedings of Machine Learning Research}, address = {Lille, France}, month = {07--09 Jul}, publisher = {PMLR}, pdf = {http://proceedings.mlr.press/v37/regier15.pdf}, url = {https://proceedings.mlr.press/v37/regier15.html}, abstract = {We present a new, fully generative model of optical telescope image sets, along with a variational procedure for inference. Each pixel intensity is treated as a Poisson random variable, with a rate parameter dependent on latent properties of stars and galaxies. Key latent properties are themselves random, with scientific prior distributions constructed from large ancillary data sets. We check our approach on synthetic images. We also run it on images from a major sky survey, where it exceeds the performance of the current state-of-the-art method for locating celestial bodies and measuring their colors.} }
Endnote
%0 Conference Paper %T Celeste: Variational inference for a generative model of astronomical images %A Jeffrey Regier %A Andrew Miller %A Jon McAuliffe %A Ryan Adams %A Matt Hoffman %A Dustin Lang %A David Schlegel %A Mr Prabhat %B Proceedings of the 32nd International Conference on Machine Learning %C Proceedings of Machine Learning Research %D 2015 %E Francis Bach %E David Blei %F pmlr-v37-regier15 %I PMLR %P 2095--2103 %U https://proceedings.mlr.press/v37/regier15.html %V 37 %X We present a new, fully generative model of optical telescope image sets, along with a variational procedure for inference. Each pixel intensity is treated as a Poisson random variable, with a rate parameter dependent on latent properties of stars and galaxies. Key latent properties are themselves random, with scientific prior distributions constructed from large ancillary data sets. We check our approach on synthetic images. We also run it on images from a major sky survey, where it exceeds the performance of the current state-of-the-art method for locating celestial bodies and measuring their colors.
RIS
TY - CPAPER TI - Celeste: Variational inference for a generative model of astronomical images AU - Jeffrey Regier AU - Andrew Miller AU - Jon McAuliffe AU - Ryan Adams AU - Matt Hoffman AU - Dustin Lang AU - David Schlegel AU - Mr Prabhat BT - Proceedings of the 32nd International Conference on Machine Learning DA - 2015/06/01 ED - Francis Bach ED - David Blei ID - pmlr-v37-regier15 PB - PMLR DP - Proceedings of Machine Learning Research VL - 37 SP - 2095 EP - 2103 L1 - http://proceedings.mlr.press/v37/regier15.pdf UR - https://proceedings.mlr.press/v37/regier15.html AB - We present a new, fully generative model of optical telescope image sets, along with a variational procedure for inference. Each pixel intensity is treated as a Poisson random variable, with a rate parameter dependent on latent properties of stars and galaxies. Key latent properties are themselves random, with scientific prior distributions constructed from large ancillary data sets. We check our approach on synthetic images. We also run it on images from a major sky survey, where it exceeds the performance of the current state-of-the-art method for locating celestial bodies and measuring their colors. ER -
APA
Regier, J., Miller, A., McAuliffe, J., Adams, R., Hoffman, M., Lang, D., Schlegel, D. & Prabhat, M.. (2015). Celeste: Variational inference for a generative model of astronomical images. Proceedings of the 32nd International Conference on Machine Learning, in Proceedings of Machine Learning Research 37:2095-2103 Available from https://proceedings.mlr.press/v37/regier15.html.

Related Material