Skip to main content Skip to navigation

Machine learning for transient science

Monday 11th - Tuesday 12th December 2023

Organisers: Mark Magee, Joe Lyman, Miika Pursiainen, Anwesha Sahu (Warwick)


In recent years, the world of transient astrophysics has undergone a paradigm shift. New wide-field and rapid surveys, including ATLAS, GOTO, and ZTF, are producing vast amounts of astronomical data every night. This has presented new opportunities for physical insights, but also new theoretical and technical challenges.

The collection of large, statistical, and homogeneous samples of transients has enabled a deeper theoretical understanding of their elusive nature. This includes detailed studies into population statistics, the rates of transients, the connection between transients and their host galaxies, cosmological applications, and more. At the same time, new and unusual transient types that confound current theoretical models are also being discovered at a rapid pace. These transients reside in previously under-populated regions of the luminosity-duration phase space and therefore could represent gaps in our understanding of stellar evolution or explosion mechanisms.

On the technical side, the sheer volume of data produced raises the question: how best do we utilise it for scientific gain? This is particularly pressing as upcoming surveys, such as LSST, will increase the number of transient discoveries even further and could become overwhelming. To that end, machine learning techniques have gained considerable attention and are now integral to various aspects of transient astrophysics, including the identification of newly discovered transients, contextual and photometric classification of transients, and even the analysis of transients through clustering and rapid modelling. As the data deluge continues, machine learning techniques will become vital to ensure data sets are fully exploited.

The aim of this informal meeting is to bring together members of the UK and wider transient community actively working on or generally interested in the broad topic of machine learning, to share ideas about techniques currently in use, current and future applications, and to foster collaboration. We invite contributed talks from members of the community, focused on their use cases and machine learning applications, and aim to have more detailed hands-on sessions centred around getting started with machine learning.

Registration of Interest

Abstract submission and registration for in-person attendance is now closed, however we welcome remote participation through Teams.Link opens in a new window

There is no fee for registration. Limited funds are available to support travel for early career researchers. Information will be provided at a later date.


  Monday 11th December 2023  
Time Title Speaker
14:00 Welcome Mark Magee (Warwick)
14:10 Sifting supernovae with deep-learned source classification Tom Killestein (Turku)
15:00 Break  
15:20 Improvements to the ATLAS real-bogus classifier Joshua Weston (QUB)
15:40 Machine Learning Applications for Images in Astronomy Anwesha Sahu (Warwick)
16:00 Discovering radio transients with machine learning and citizen science Alex Andersson (Oxford)
16:20 Break  
16:40 Spectral modelling of supernovae using emulation Mark Magee (Warwick)
17:00 Discussion  
18:00 End  
  Tuesday 12th December 2023  
Time Title Speaker
09:30 Augmenting supernova training sets using Generative Adversarial Networks Matt Grayling (Cambridge)
10:20 GausSN: Bayesian Time Delay Estimation for Strongly Lensed Supernovae Erin Hayes (Cambridge)
10:40 Multi-Modal Insights for Early Photometric Classification of Supernovae in the Era of LSST Alexander Gagliano (Harvard/MIT)




Just because you can doesn't mean you should Heloise Stevance (Oxford)





Abstracts can be found hereLink opens in a new window. The meeting will also be available online with TeamsLink opens in a new window. All times are given in GMT.


Millburn House, A.028

Please follow the link to the interactive campus mapLink opens in a new window.



Mark Magee
Department of Physics
University of Warwick