Lessons Learned from the 1st ARIEL Machine Learning Challenge - Correcting Transiting Exoplanet Light Curves for Stellar Spots

Aug 5, 2026

The last decade has driven rapid progress in exoplanet discovery and characterization, while major challenges remain that can be addressed with machine learning. Transit photometry, a leading method for detecting exoplanets and deriving properties, is strongly affected by stellar spots, which current workflows correct manually or exclude. This work presents initial steps toward fully automated, accurate extraction of transit depths under spot contamination, detailing a range of methods.

LanguageEnglish

CategoryResearch & Report

FormatPDF

Pages18

Reading Time45 min

AccessFree to download

Abstract

Introduction

ORCA-Online Research @Cardiff This is an Open Access document downloaded from ORCA,Cardiff University'sinstitutional repository:https://orca.cardiff.ac.uk/id/eprint/163802/ This is the author's version of a work that was submitted to /accepted forpublication. Citation for final published version: Nikolaou,Nikolaos,Waldmann,Ingo P.,Tsiaras,Angelos,Morvan,Mario,Edwards,Billy,Yip,Kai Hou,Thompson,Alexandra,Tinetti,Giovanna,Sarkar,Subhajit,Dawson,James M.,Borisov,Vadim,Kasneci,Gjergji,Petkovic,Matej,Stepisnik,Tomaz,Al-Ubaidi,Tarek,Bailey,Rachel Louise,Granitzer,Michael,Julka,Sahib,Kern,Roman,Ofner,Patrick,Wagner,Stefan,Heppe,Lukas,Bunse,Mirko,Morik,Katharina and Simoes,Luís F.2023.Lessons learned from the 1st Ariel MachineLearning Challenge:Correcting transiting exoplanet light curves for stellar spots.RAS Techniques and Instruments 10.1093/rasti/rzad050 Publishers page:http://dx.doi.org/10.1093/rasti/rzad050 Please note: Changes made as a result of publishing processes such as copy-e diting,formattingand page numbers may not be reflected in this version.For the definitive version ofthis publication,please refer to the published source.You are advised to consult thepublisher's version if you wish to cite this paper. This version is being made available in accordance with publisher policies.Seehttp://orca.cf.ac.uk/policies.html for usage policies.Copyright and moral rights forpublications made available in ORCA are retained by the copyright holders. # Lessons Learned from the 1st ARIEL Machine Learning Challenge:Correcting Transiting Exoplanet Light Curves for Stellar Spots Nikolaos Nikolaou,*Ingo P.Waldmann,Angelos Tsiaras,2,1 Mario Morvan,Billy Edwards,Kai Hou Yip,¹Alexandra Thompson,¹Giovanna Tinetti,¹Subhajit Sarkar,3James M.Dawson,³Vadim Borisov,⁴Gjergji Kasneci,4 Matej Petkovic,5 Tomaž Stepišnik,5 Tarek Al-Ubaidi,6,7Rachel Louise Bailey,?Michael Granitzer,8 Sahib Julka,8Roman Kern,9 Patrick Ofner,9Stefan Wagner,10 Lukas Heppe,¹Mirko Bunse,¹Katharina Morik¹¹and Luís F.Simoes¹2 Department of Physics and Astronomy,University College London,Gower Street,London,WCIE6BT,UK2INAF—Osservatorio Astrofisico diArcetri,Largo E.Fermi 5,I-50125 Firenze,Italy3School ofPhysics and Astronomy,Cardif University,The Parade,Cardif,CF243AA,UK4Department of Computer Science,University of Tuebingen,Tuebingen,Germany5Jožef Stefan Institute,Ljubljana,Slovenia6DCCS GmbH,Austria ⁷Space Research Institute,Austrian Academy of Sciences,Austria8Chair of Data Science,University of Passau,Germany 9Know-Center GmbH,Research Center for Data-Driven Business &Big Data Analytics,Austria10Commission for Astronomy,Austrian Academy of Sciences,Graz,Austria llLamarr hnstitute for Machine Leaming and Arifcial Inteligence,TU Dortmund University,Germany12MLAnalytics,Lisbon,Portugal Accepted XXX.Received YYY;in original form ZZZ ## ABSTRACT The last decade has witnessed a rapid growth of the field of exoplanet discovery and characterisation.However,several bigchallenges remain,many of which could be addressed using machine learning methodology.For instance,the most prolificmethod for detecting exoplanets and inferring several of their characteristics,transit photometry,is very sensitive to the presenceof stellar spots.The current practice in the literature is to identify the effects of spots visually and correct for them manually ordiscard the affected data.This paper explores a first step towards fully automating the efficient and precise derivation of transitdepthsfrom transit light curves in the presence of stellar spots.The primary focus of the paper is to present in detail a diversearsenal of methods for doing so.The methods and results we present were obtained in the context of the 1st Machine LearningChallenge organized for the European Space Agency's upcoming Ariel mission.We first present the problem,the simulatedAriel-like data and outline the Challenge while identifying best practices for organizing similar challenges in the futur

FAQs

Why are stellar spots a problem for transit photometry?

Stellar spots introduce noise and bias the measured transit depth, impacting the derived transmission spectrum and other light-curve parameters such as limb-darkening and mid-transit times.

What is the main goal of the paper?

To move toward fully automated and precise derivation of transit depths from transit light curves in the presence of stellar spots, using a diverse set of methods.

How were the methods evaluated in this study?

The methods and results were developed in the context of the 1st Machine Learning Challenge for the European Space Agency’s upcoming Ariel mission, and the paper reports top-team solutions and implications.

Aug 5, 2026

Upload by Olivia Brown