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

Killestein, T. L.; 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.; Aukkaravittayapun, S.; Awiphan, S.; Burhanudin, U.; Chote, P.; Chrimes, A.; Daw, E.; Duffy, C.; Eyles-Ferris, R.; Gompertz, B.; Heikkilä, T.; Irawati, P.; Kennedy, M. R.; 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, P. A.; Tooke, S.; Wiersema, K.; Williams, S. C.
Bibliographical reference

Monthly Notices of the Royal Astronomical Society

Advertised on:
5
2021
Description
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.
Related projects
Representación de la variable cataclísmica SS Cygni (Chris Moran)
Binary Stars

The study of binary stars is essential to stellar astrophysics. A large number of stars form and evolve within binary systems. Therefore, their study is fundamental to understand stellar and galactic evolution. Particularly relevant is that binary systems are still the best source of precise stellar mass and radius measurements. Research lines

Pablo
Rodríguez Gil
Black hole in outburst
Black holes, neutron stars, white dwarfs and their local environment

Accreting black-holes and neutron stars in X-ray binaries provide an ideal laboratory for exploring the physics of compact objects, yielding not only confirmation of the existence of stellar mass black holes via dynamical mass measurements, but also the best opportunity for probing high-gravity environments and the physics of accretion; the most

Teodoro
Muñoz Darias
Projects' name image
Exoplanets and Astrobiology

The search for life in the universe has been driven by recent discoveries of planets around other stars (known as exoplanets), becoming one of the most active fields in modern astrophysics. The growing number of new exoplanets discovered in recent years and the recent advance on the study of their atmospheres are not only providing new valuable

Enric
Pallé Bago