Home > Research > Publications & Outputs > Transient-optimized real-bogus classification w...

Research

Associated organisational unit

Links

Text available via DOI:

View graph of relations

Transient-optimized real-bogus classification with Bayesian convolutional neural networks – sifting the GOTO candidate stream

Research output: Contribution to Journal/Magazine › Journal article › peer-review

Published

Standard

Transient-optimized real-bogus classification with Bayesian convolutional neural networks – sifting the GOTO candidate stream. / Killestein, Thomas; Lyman, Joe; Steeghs, D et al.
In: Monthly Notices of the Royal Astronomical Society, Vol. 503, No. 4, 30.06.2021, p. 4838-4854.

Research output: Contribution to Journal/Magazine › Journal article › peer-review

Harvard

Killestein, T, Lyman, J, Steeghs, D, Ackley, K, Dyer, MJ, Ulaczyk, K, Cutter, R, Mong, Y-L, Galloway, DK, Dhillon, V, O’Brien, P, Ramsay, G, Poshyachinda, S, Kotak, R, Breton, RP, Nuttall, LK, Pallé, E, Pollacco, D, Thrane, E, Aukkaravittayapun, S, Awiphan, S, Burhanudin, U, Chote, P, Chrimes, A, Daw, E, Duffy, C, Eyles-Ferris, R, Gompertz, B, Heikkilä, T, Irawati, P, Kennedy, MR, Levan, A, Littlefair, S, Makrygianni, L, Mata Sánchez, D, Mattila, S, Maund, J, McCormac, J, Mkrtichian, D, Mullaney, J, Rol, E, Sawangwit, U, Stanway, E, Starling, R, Strøm, PA, Tooke, S, Wiersema, K & Williams, SC 2021, 'Transient-optimized real-bogus classification with Bayesian convolutional neural networks – sifting the GOTO candidate stream', Monthly Notices of the Royal Astronomical Society, vol. 503, no. 4, pp. 4838-4854. https://doi.org/10.1093/mnras/stab633

APA

Killestein, T., Lyman, J., Steeghs, D., Ackley, K., Dyer, M. J., Ulaczyk, K., Cutter, R., Mong, Y.-L., Galloway, D. K., Dhillon, V., O’Brien, P., Ramsay, G., Poshyachinda, S., Kotak, R., Breton, R. P., Nuttall, L. K., Pallé, E., Pollacco, D., Thrane, E., ... Williams, S. C. (2021). Transient-optimized real-bogus classification with Bayesian convolutional neural networks – sifting the GOTO candidate stream. Monthly Notices of the Royal Astronomical Society, 503(4), 4838-4854. https://doi.org/10.1093/mnras/stab633

Vancouver

Killestein T, Lyman J, Steeghs D, Ackley K, Dyer MJ, Ulaczyk K et al. Transient-optimized real-bogus classification with Bayesian convolutional neural networks – sifting the GOTO candidate stream. Monthly Notices of the Royal Astronomical Society. 2021 Jun 30;503(4):4838-4854. Epub 2021 Mar 15. doi: 10.1093/mnras/stab633

Author

Killestein, Thomas ; Lyman, Joe ; Steeghs, D et al. / Transient-optimized real-bogus classification with Bayesian convolutional neural networks – sifting the GOTO candidate stream. In: Monthly Notices of the Royal Astronomical Society. 2021 ; Vol. 503, No. 4. pp. 4838-4854.

Bibtex

@article{59a42261117f4df7982e75c0bf13f127,
title = "Transient-optimized real-bogus classification with Bayesian convolutional neural networks – sifting the GOTO candidate stream",
abstract = "Large-scale sky surveys have played a transformative role in our understanding of astrophysical transients, only made possible by increasingly powerful machine learning-based filtering to accurately sift through the vast quantities of incoming data generated. In this paper, we present a new real-bogus classifier based on a Bayesian convolutional neural network that provides nuanced, uncertainty-aware classification of transient candidates in difference imaging, and demonstrate its application to the datastream from the GOTO wide-field optical survey. Not only are candidates assigned a well-calibrated probability of being real, but also an associated confidence that can be used to prioritize human vetting efforts and inform future model optimization via active learning. To fully realize the potential of this architecture, we present a fully automated training set generation method which requires no human labelling, incorporating a novel data-driven augmentation method to significantly improve the recovery of faint and nuclear transient sources. We achieve competitive classification accuracy (FPR and FNR both below 1 per cent) compared against classifiers trained with fully human-labelled data sets, while being significantly quicker and less labour-intensive to build. This data-driven approach is uniquely scalable to the upcoming challenges and data needs of next-generation transient surveys. We make our data generation and model training codes available to the community.",
author = "Thomas Killestein and Joe Lyman and D Steeghs and K Ackley and Dyer, {Martin J} and K Ulaczyk and R. Cutter and Y-L Mong and Galloway, {D K} and Vik Dhillon and P O{\textquoteright}Brien and G Ramsay and S Poshyachinda and R Kotak and Breton, {Rene P} and Nuttall, {L K} and E Pall{\'e} and D Pollacco and E Thrane and S Aukkaravittayapun and S Awiphan and U Burhanudin and P Chote and A Chrimes and E Daw and Christopher Duffy and R Eyles-Ferris and Benjamin Gompertz and T Heikkil{\"a} and P Irawati and Kennedy, {Mark R} and A Levan and S Littlefair and L Makrygianni and D Mata S{\'a}nchez and S Mattila and Justyn Maund and J McCormac and D Mkrtichian and J Mullaney and E Rol and U Sawangwit and Elizabeth Stanway and Rhaana Starling and Str{\o}m, {P A} and S Tooke and K Wiersema and Williams, {Steven C.}",
year = "2021",
month = jun,
day = "30",
doi = "10.1093/mnras/stab633",
language = "English",
volume = "503",
pages = "4838--4854",
journal = "Monthly Notices of the Royal Astronomical Society",
issn = "0035-8711",
publisher = "OXFORD UNIV PRESS",
number = "4",

}

RIS

TY - JOUR

T1 - Transient-optimized real-bogus classification with Bayesian convolutional neural networks – sifting the GOTO candidate stream

AU - Killestein, Thomas

AU - Lyman, Joe

AU - Steeghs, D

AU - Ackley, K

AU - Dyer, Martin J

AU - Ulaczyk, K

AU - Cutter, R.

AU - Mong, Y-L

AU - Galloway, D K

AU - Dhillon, Vik

AU - O’Brien, P

AU - Ramsay, G

AU - Poshyachinda, S

AU - Kotak, R

AU - Breton, Rene P

AU - Nuttall, L K

AU - Pallé, E

AU - Pollacco, D

AU - Thrane, E

AU - Aukkaravittayapun, S

AU - Awiphan, S

AU - Burhanudin, U

AU - Chote, P

AU - Chrimes, A

AU - Daw, E

AU - Duffy, Christopher

AU - Eyles-Ferris, R

AU - Gompertz, Benjamin

AU - Heikkilä, T

AU - Irawati, P

AU - Kennedy, Mark R

AU - Levan, A

AU - Littlefair, S

AU - Makrygianni, L

AU - Mata Sánchez, D

AU - Mattila, S

AU - Maund, Justyn

AU - McCormac, J

AU - Mkrtichian, D

AU - Mullaney, J

AU - Rol, E

AU - Sawangwit, U

AU - Stanway, Elizabeth

AU - Starling, Rhaana

AU - Strøm, P A

AU - Tooke, S

AU - Wiersema, K

AU - Williams, Steven C.

PY - 2021/6/30

Y1 - 2021/6/30

N2 - Large-scale sky surveys have played a transformative role in our understanding of astrophysical transients, only made possible by increasingly powerful machine learning-based filtering to accurately sift through the vast quantities of incoming data generated. In this paper, we present a new real-bogus classifier based on a Bayesian convolutional neural network that provides nuanced, uncertainty-aware classification of transient candidates in difference imaging, and demonstrate its application to the datastream from the GOTO wide-field optical survey. Not only are candidates assigned a well-calibrated probability of being real, but also an associated confidence that can be used to prioritize human vetting efforts and inform future model optimization via active learning. To fully realize the potential of this architecture, we present a fully automated training set generation method which requires no human labelling, incorporating a novel data-driven augmentation method to significantly improve the recovery of faint and nuclear transient sources. We achieve competitive classification accuracy (FPR and FNR both below 1 per cent) compared against classifiers trained with fully human-labelled data sets, while being significantly quicker and less labour-intensive to build. This data-driven approach is uniquely scalable to the upcoming challenges and data needs of next-generation transient surveys. We make our data generation and model training codes available to the community.

AB - Large-scale sky surveys have played a transformative role in our understanding of astrophysical transients, only made possible by increasingly powerful machine learning-based filtering to accurately sift through the vast quantities of incoming data generated. In this paper, we present a new real-bogus classifier based on a Bayesian convolutional neural network that provides nuanced, uncertainty-aware classification of transient candidates in difference imaging, and demonstrate its application to the datastream from the GOTO wide-field optical survey. Not only are candidates assigned a well-calibrated probability of being real, but also an associated confidence that can be used to prioritize human vetting efforts and inform future model optimization via active learning. To fully realize the potential of this architecture, we present a fully automated training set generation method which requires no human labelling, incorporating a novel data-driven augmentation method to significantly improve the recovery of faint and nuclear transient sources. We achieve competitive classification accuracy (FPR and FNR both below 1 per cent) compared against classifiers trained with fully human-labelled data sets, while being significantly quicker and less labour-intensive to build. This data-driven approach is uniquely scalable to the upcoming challenges and data needs of next-generation transient surveys. We make our data generation and model training codes available to the community.

U2 - 10.1093/mnras/stab633

DO - 10.1093/mnras/stab633

M3 - Journal article

VL - 503

SP - 4838

EP - 4854

JO - Monthly Notices of the Royal Astronomical Society

JF - Monthly Notices of the Royal Astronomical Society

SN - 0035-8711

IS - 4

ER -