Supervised Machine Learning Methods with Uncertainty Quantification for Exoplanet Atmospheric Retrievals from Transmission Spectroscopy

ABSTRACT Standard Bayesian retrievals for exoplanet atmospheric parameters from transmission spectroscopy, while well understood and widely used, are generally computationally expensive. In the era of the JWST and other upcoming observatories, machine learning approaches have emerged as viable alternatives that are both efficient and robust. In this paper we present a systematic study of several existing machine learning regression techniques and compare their performance for retrieving exoplanet atmospheric parameters from transmission spectra. The regression methods tested here include partial least squares (PLS), support vector machines (SVM), k nearest neighbors (KNN), decision trees (DT), random forests (RF), voting (VOTE), stacking (STACK), and extreme gradient boosting (XGB). We also investigate the impact of different preprocessing methods of the training data on the model performance. We quantify the model uncertainties across the entire dynamical range of planetary parameters. The best performing combination of ML model and preprocessing scheme is validated on a the case study of JWST observation of WASP-39b. INTRODUCTION Over the last three decades, the study of extrasolar system planets has shifted from discovery to inference with particular interest in the characterization of their chemical compositions and temperature profiles. The chemical inventory of an exoplanet atmosphere is impacted by the planet formation processes, evolutionary modifications, and its interactions with the local space environment, thus allowing us to place constraints on the existing evolutionary models from the retrieved atmospheric composition. Transit spectroscopy is currently the most widely used observational technique to study the chemical composition of transiting exoplanets (Schneider 1994; Charbonneau et al. 2000). During transit, the planet atmosphere is observed in transmitted light when a planet passes in front of its host star, i.e., the primary eclipse, and in emitted and/or reflected light when a planet travels behind its host star, referred to as the secondary eclipse.