Studies on the use of ML algorithms in exoplanet spectra. We describe the purpose and application of the algorithms used in each study. Additionally, we list the exoplanet regimes each algorithm focuses on and indicate whether they were trained or tested with C/O ratios plus metallicity (represented in the column C/O), or the most common free chemistry molecules.
| Reference . | ML method(s) . | Targets . | C/O . | H|$_2$|O . | CO|$_2$| . | CO . | CH|$_4$| . | NH|$_3$| . | HCN . | O|$_3$| . |
|---|---|---|---|---|---|---|---|---|---|---|
| Gebhard et al. (2024) | Convolutional neural networks and multilayer perceptrons (regression) | Hot Jupiters and Earth-like planets | ||||||||
| Ardévol Martínez et al. (2024) | Sequential neural posterior estimation with normalizing flows (regression) | Wide range of planets, brown dwarfs | ||||||||
| Forestano et al. (2023) | Local outlier factor, one-class support vector machine (unsupervised machine learning, outlier detection) | Hot Jupiters | ||||||||
| Vasist et al. (2023) | Neural posterior estimation with normalizing flows (regression) | Gas giants | ||||||||
| Ardévol Martínez et al. (2022) | Convolutional neural networks (regression) | Gas giants | ||||||||
| Himes et al. (2022) | Convolutional neural networks (regression) | Hot Jupiters | ||||||||
| Matchev, Matcheva & Roman (2022a) | k-means, PCA and ISOMAP (Unsupervised) | Hot Jupiters | ||||||||
| Matchev, Matcheva & Roman (2022b) | Symbolic regression | Hot Jupiters | ||||||||
| Munsaket et al. (2021) | Random forest (supervised ML, regression) | Hot Jupiters | ||||||||
| Yip et al. (2021) | Multilayer perceptrons, convolutional neural networks and long short-term memory networks (regression) | Wide range of planets | ||||||||
| Guzmán-Mesa et al. (2020) | Random forest (regression) | Warm Neptunes | ||||||||
| Hayes et al. (2020) | K-means, principal component analysis (unsupervised classification) | Jupiter-like planets | ||||||||
| Nixon & Madhusudhan (2020) | Random forest (regression) | Hot Jupiters | ||||||||
| Cobb et al. (2019) | Bayesian neural network (regression) | Hot Jupiters | ||||||||
| Soboczenski et al. (2018) | Convolutional neural networks with Monte Carlo dropout (regression) | Rocky terrestrial exoplanets | ||||||||
| Marquez-Neila et al. (2018) | Random forest (regression) | Hot Jupiters | ||||||||
| Zingales & Waldmann (2018) | Generative adversarial networks (regression) | Hot Jupiters | ||||||||
| Waldmann (2016) | Deep-belief networks (classification) | Wide range of planets |
| Reference . | ML method(s) . | Targets . | C/O . | H|$_2$|O . | CO|$_2$| . | CO . | CH|$_4$| . | NH|$_3$| . | HCN . | O|$_3$| . |
|---|---|---|---|---|---|---|---|---|---|---|
| Gebhard et al. (2024) | Convolutional neural networks and multilayer perceptrons (regression) | Hot Jupiters and Earth-like planets | ||||||||
| Ardévol Martínez et al. (2024) | Sequential neural posterior estimation with normalizing flows (regression) | Wide range of planets, brown dwarfs | ||||||||
| Forestano et al. (2023) | Local outlier factor, one-class support vector machine (unsupervised machine learning, outlier detection) | Hot Jupiters | ||||||||
| Vasist et al. (2023) | Neural posterior estimation with normalizing flows (regression) | Gas giants | ||||||||
| Ardévol Martínez et al. (2022) | Convolutional neural networks (regression) | Gas giants | ||||||||
| Himes et al. (2022) | Convolutional neural networks (regression) | Hot Jupiters | ||||||||
| Matchev, Matcheva & Roman (2022a) | k-means, PCA and ISOMAP (Unsupervised) | Hot Jupiters | ||||||||
| Matchev, Matcheva & Roman (2022b) | Symbolic regression | Hot Jupiters | ||||||||
| Munsaket et al. (2021) | Random forest (supervised ML, regression) | Hot Jupiters | ||||||||
| Yip et al. (2021) | Multilayer perceptrons, convolutional neural networks and long short-term memory networks (regression) | Wide range of planets | ||||||||
| Guzmán-Mesa et al. (2020) | Random forest (regression) | Warm Neptunes | ||||||||
| Hayes et al. (2020) | K-means, principal component analysis (unsupervised classification) | Jupiter-like planets | ||||||||
| Nixon & Madhusudhan (2020) | Random forest (regression) | Hot Jupiters | ||||||||
| Cobb et al. (2019) | Bayesian neural network (regression) | Hot Jupiters | ||||||||
| Soboczenski et al. (2018) | Convolutional neural networks with Monte Carlo dropout (regression) | Rocky terrestrial exoplanets | ||||||||
| Marquez-Neila et al. (2018) | Random forest (regression) | Hot Jupiters | ||||||||
| Zingales & Waldmann (2018) | Generative adversarial networks (regression) | Hot Jupiters | ||||||||
| Waldmann (2016) | Deep-belief networks (classification) | Wide range of planets |
Studies on the use of ML algorithms in exoplanet spectra. We describe the purpose and application of the algorithms used in each study. Additionally, we list the exoplanet regimes each algorithm focuses on and indicate whether they were trained or tested with C/O ratios plus metallicity (represented in the column C/O), or the most common free chemistry molecules.
| Reference . | ML method(s) . | Targets . | C/O . | H|$_2$|O . | CO|$_2$| . | CO . | CH|$_4$| . | NH|$_3$| . | HCN . | O|$_3$| . |
|---|---|---|---|---|---|---|---|---|---|---|
| Gebhard et al. (2024) | Convolutional neural networks and multilayer perceptrons (regression) | Hot Jupiters and Earth-like planets | ||||||||
| Ardévol Martínez et al. (2024) | Sequential neural posterior estimation with normalizing flows (regression) | Wide range of planets, brown dwarfs | ||||||||
| Forestano et al. (2023) | Local outlier factor, one-class support vector machine (unsupervised machine learning, outlier detection) | Hot Jupiters | ||||||||
| Vasist et al. (2023) | Neural posterior estimation with normalizing flows (regression) | Gas giants | ||||||||
| Ardévol Martínez et al. (2022) | Convolutional neural networks (regression) | Gas giants | ||||||||
| Himes et al. (2022) | Convolutional neural networks (regression) | Hot Jupiters | ||||||||
| Matchev, Matcheva & Roman (2022a) | k-means, PCA and ISOMAP (Unsupervised) | Hot Jupiters | ||||||||
| Matchev, Matcheva & Roman (2022b) | Symbolic regression | Hot Jupiters | ||||||||
| Munsaket et al. (2021) | Random forest (supervised ML, regression) | Hot Jupiters | ||||||||
| Yip et al. (2021) | Multilayer perceptrons, convolutional neural networks and long short-term memory networks (regression) | Wide range of planets | ||||||||
| Guzmán-Mesa et al. (2020) | Random forest (regression) | Warm Neptunes | ||||||||
| Hayes et al. (2020) | K-means, principal component analysis (unsupervised classification) | Jupiter-like planets | ||||||||
| Nixon & Madhusudhan (2020) | Random forest (regression) | Hot Jupiters | ||||||||
| Cobb et al. (2019) | Bayesian neural network (regression) | Hot Jupiters | ||||||||
| Soboczenski et al. (2018) | Convolutional neural networks with Monte Carlo dropout (regression) | Rocky terrestrial exoplanets | ||||||||
| Marquez-Neila et al. (2018) | Random forest (regression) | Hot Jupiters | ||||||||
| Zingales & Waldmann (2018) | Generative adversarial networks (regression) | Hot Jupiters | ||||||||
| Waldmann (2016) | Deep-belief networks (classification) | Wide range of planets |
| Reference . | ML method(s) . | Targets . | C/O . | H|$_2$|O . | CO|$_2$| . | CO . | CH|$_4$| . | NH|$_3$| . | HCN . | O|$_3$| . |
|---|---|---|---|---|---|---|---|---|---|---|
| Gebhard et al. (2024) | Convolutional neural networks and multilayer perceptrons (regression) | Hot Jupiters and Earth-like planets | ||||||||
| Ardévol Martínez et al. (2024) | Sequential neural posterior estimation with normalizing flows (regression) | Wide range of planets, brown dwarfs | ||||||||
| Forestano et al. (2023) | Local outlier factor, one-class support vector machine (unsupervised machine learning, outlier detection) | Hot Jupiters | ||||||||
| Vasist et al. (2023) | Neural posterior estimation with normalizing flows (regression) | Gas giants | ||||||||
| Ardévol Martínez et al. (2022) | Convolutional neural networks (regression) | Gas giants | ||||||||
| Himes et al. (2022) | Convolutional neural networks (regression) | Hot Jupiters | ||||||||
| Matchev, Matcheva & Roman (2022a) | k-means, PCA and ISOMAP (Unsupervised) | Hot Jupiters | ||||||||
| Matchev, Matcheva & Roman (2022b) | Symbolic regression | Hot Jupiters | ||||||||
| Munsaket et al. (2021) | Random forest (supervised ML, regression) | Hot Jupiters | ||||||||
| Yip et al. (2021) | Multilayer perceptrons, convolutional neural networks and long short-term memory networks (regression) | Wide range of planets | ||||||||
| Guzmán-Mesa et al. (2020) | Random forest (regression) | Warm Neptunes | ||||||||
| Hayes et al. (2020) | K-means, principal component analysis (unsupervised classification) | Jupiter-like planets | ||||||||
| Nixon & Madhusudhan (2020) | Random forest (regression) | Hot Jupiters | ||||||||
| Cobb et al. (2019) | Bayesian neural network (regression) | Hot Jupiters | ||||||||
| Soboczenski et al. (2018) | Convolutional neural networks with Monte Carlo dropout (regression) | Rocky terrestrial exoplanets | ||||||||
| Marquez-Neila et al. (2018) | Random forest (regression) | Hot Jupiters | ||||||||
| Zingales & Waldmann (2018) | Generative adversarial networks (regression) | Hot Jupiters | ||||||||
| Waldmann (2016) | Deep-belief networks (classification) | Wide range of planets |
This PDF is available to Subscribers Only
View Article Abstract & Purchase OptionsFor full access to this pdf, sign in to an existing account, or purchase an annual subscription.
