Skip to main content
SearchLoginLogin or Signup

Characterising Exoplanets with Machine Learning: Lesson Learnt from the Ariel Data Challenge 2022 and 2023

Presentation #109.02 in the session Open Science and Data Tools (Oral Presentation)

Published onOct 23, 2023
Characterising Exoplanets with Machine Learning: Lesson Learnt from the Ariel Data Challenge 2022 and 2023

Current endeavours in exoplanet characterisation rely on atmospheric retrieval to quantify crucial physical properties of remote exoplanets from observations. However, the scalability and efficiency of the technique are under strain with increasing spectroscopic resolution and forward model complexity. The situation becomes more acute with the recent launch of the James Webb Space Telescope and the Ariel Space Mission. Machine learning (ML) techniques lands itself as a promising alternative. Developing techniques of this kind is an inter-disciplinary task, one that requires domain knowledge of the field, access to relevant tools and expert insights on the capability and limitations of current ML models. These stringent requirements have so far limited the developments of ML in the field to a few isolated initiatives. Accepted as part of the competition track in NeurIPS 2022 and ECML-PKDD 2023, the goal of the Ariel Data Challenge is to open up the problem to the global community and identify reliable and scalable method(s) to perform planetary characterisation. Participants were given 2-4 months to train a model that can provide the approximate conditional distribution of key planetary properties. The data challenge is positively received by the community and has attracted 250+ participants from 51 countries around the globe to participate. In this talk I will focus on the outcomes of the data challenges. In particular, I will talk about 1.) the winning solutions 2.) the publicly available ABC database and 3.) the impact of dataset shift on model performance and its relevance to a space mission.

Comments
0
comment
No comments here