Pith. sign in

REVIEW 4 major objections 3 minor 1 cited by

Towards the Habitable Worlds Observatory: 1D CNN Retrieval of Reflection Spectra from Evolving Earth Analogs

T0 review · 4 major / 3 minor · reviewed 2026-08-06 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗

Pith's one-line read A 1D CNN retrieves six gas abundances and planet parameters from reflected-light spectra in seconds.

desk verdict A solid synthetic retrieval demo for Earth analogs, but the abstract's 'mission-ready' label overclaims what in-sample CNN validation can support. read the letter →

arxiv 2508.00076 v1 pith:EZQV3F5U submitted 2025-07-31 astro-ph.EP

Sarah G. A. Barbosa , Raissa Estrela , Paulo C. F. da Silva Filho , Daniel B. de Freitas This is my paper
classification astro-ph.EP
keywords exoplanetatmospheresatmosphericretrievalconvolutionalneuralnetworkreflected-lightspectroscopybiosignaturesEarthanalogsHabitableWorldsObservatoryMonteCarloDropout
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

The paper tries to establish that a one-dimensional convolutional neural network (1D CNN) can replace slow Bayesian retrieval for reflected-light spectra of Earth-like exoplanets, retrieving six molecular abundances plus radius, surface gravity, surface pressure, and temperature simultaneously. The network is trained on more than a million synthetic, noise-injected spectra simulating Archean, Proterozoic, and Modern Earth as seen by the LUVOIR-B and HabEx/SS instrument concepts, and tested on unseen spectra. It recovers CH$_4$ and CO$_2$ in Archean atmospheres and O$_2$ and O$_3$ in Modern cases, and outputs near-zero abundances for gases that are truly absent. The authors conclude that the network is a mission-ready retrieval engine for the Habitable Worlds Observatory, able to process direct-imaging spectra on an operational cadence with uncertainties from Monte Carlo Dropout.

What carries the argument

The load-bearing object is the 1D CNN itself: a convolutional architecture that takes a binned reflected-light spectrum as input and outputs estimates for six molecular abundances, including O$_2$ and O$_3$, plus radius, gravity, surface pressure, and temperature. It is trained on over one million synthetic spectra generated with a forward radiative-transfer model of Archean, Proterozoic, and Modern Earth atmospheres, with noise injected to mimic LUVOIR-B and HabEx/SS observations. Uncertainty estimates come from Monte Carlo Dropout, which runs thousands of stochastic forward passes through the trained network in seconds. Integrated Gradients is used to attribute the network's predictions to wavelength regions, and credibility curves are used to map detectability as a function of stellar type and distance.

What would settle it

Take the trained network and feed it a real Earthshine spectrum of the present-day Earth, or a high-fidelity empirical spectrum from a solar system analog, and compare its retrieved O$_2$, O$_3$, radius, gravity, surface pressure, and temperature against known values. Disagreement beyond the reported credible intervals, or disagreement with an independent Bayesian retrieval run on the same spectrum, would show that the mission-ready claim does not transfer to real data.

Watch

Extended reading notes

Core claim

The central claim is that a single 1D CNN, trained on synthetic reflected-light spectra of Earth analogs, accurately infers atmospheric composition and bulk parameters from direct-imaging observations in seconds, with uncertainties, and does so without inventing biosignatures where none exist. On the paper's own evidence, the network performs best where spectral features are strongest: CH$_4$ and CO$_2$ in Archean cases, and O$_2$ and O$_3$ in Modern cases. Integrated Gradients attribution shows the model keys on physically meaningful absorption bands, the Fraunhofer A band for O$_2$ and the Hartley-Huggins band for O$_3$, and credibility curves show O$_3$ stays retrievable across many stellar types and distances while O$_2$ is detectable out to 12 pc around FG stars. The paper presents these results as moving CNNs from proof of concept to a mission-ready retrieval engine for HWO.

Load-bearing premise

The weakest link is that the synthetic forward model and injected noise used to create the million training spectra match real HWO observations closely enough; if that match fails, the network's accuracies, uncertainties, and detectability distances could all be wrong even though it scores perfectly on its own test set.

Editorial extensions

If this is right

  • Retrievals for HWO targets could be completed in seconds, allowing rapid triage of dozens of candidate Earth analogs before any expensive follow-up.
  • Simultaneous inference of O$_2$/O$_3$ with other gases and bulk parameters would let mission planners rank targets by biosignature confidence directly from reflected-light spectra.
  • Near-zero abundance outputs for absent species, if they persist on real data, would suppress the most dangerous false-positive biosignature claims.
  • The reported detectability distances, with O$_2$ out to 12 pc around FG stars and O$_3$ across a wider range, provide concrete input for designing HWO target lists and exposure times.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • Beyond the paper, a natural test is to run the same network on Earthshine spectra of the real Earth, whose abundances and parameters are known, to see whether the synthetic-to-real generalization actually holds.
  • Because the network can only interpolate within the chemistry and geometry of its training set, real planets with hazes, clouds, or unknown gas mixtures may fall outside the training manifold; out-of-distribution tests would reveal whether the reported uncertainties remain honest.
  • The same architecture could be applied to thermal-emission spectra or to non-Earth analog compositions if the training grid were expanded, making the retrieval engine a candidate pipeline for a broader set of HWO target classes.
  • Combining the credibility curves with stellar occurrence rates would turn this retrieval engine into an observing-strategy tool that predicts how many detectable Earth analogs HWO might find.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

4 major / 3 minor

Summary. The paper presents a one-dimensional convolutional neural network (1D CNN) trained on over one million synthetic, noise-injected reflected-light spectra of Archean, Proterozoic, and Modern Earth analogs, as observed by the proposed LUVOIR-B and HabEx/SS instruments. The model simultaneously infers six molecular abundances (including O2 and O3) along with radius, gravity, surface pressure, and temperature, with uncertainties obtained via Monte Carlo Dropout. The authors report accurate recovery of CH4 and CO2 in Archean atmospheres and O2 and O3 in Modern cases, avoidance of false positives for absent species, physically meaningful attributions through Integrated Gradients, and credibility curves indicating detectability of O3 across a wide range of stellar types and distances and O2 out to 12 pc around FG stars. Based on these tests, the paper claims that the CNN is a 'mission-ready retrieval engine' for the Habitable Worlds Observatory.

Significance. If the central claim is established, this work would be a meaningful step toward fast atmospheric retrieval for direct-imaging missions, addressing a real computational bottleneck in exoplanet characterization. The paper has several strengths: a large and diverse training set spanning three Earth epochs, explicit treatment of uncertainties through Monte Carlo Dropout, interpretability analysis via Integrated Gradients with reference to specific spectral bands, and an attempt to quantify detectability as a function of stellar type and distance. These elements go beyond many proof-of-concept neural retrieval studies. However, the evidence presented in the abstract and the readable portions of the manuscript is entirely in-sample: the network is trained and tested on spectra generated by the same forward model, with no independent cross-validation, no application to real observations (e.g., Earthshine spectra), and no quantitative metrics such as bias, scatter, coverage probability, or false-positive rates at defined detection thresholds. The 'mission-ready' claim therefore rests on an unverified synthetic-to-real transfer assumption.

major comments (4)
  1. [Abstract (validation)] The central claim that the CNN is 'mission-ready' is not supported by the reported validation, which is entirely internal to a single synthetic forward-model pipeline. The network is trained on one million spectra and tested on unseen spectra drawn from the same generator; this establishes in-sample generalization but not the ability to retrieve real HWO observations. The authors need to add an external validation step, such as comparing retrievals against an independent radiative transfer code with different assumptions (e.g., different cloud treatment, line lists, or stellar contamination models), and/or applying the network to real disk-integrated Earthshine spectra, before claiming operational readiness. At minimum, the paper must include a quantitative failure analysis showing how retrieval bias and uncertainty calibration degrade when the input spectra depart from the training distribution.
  2. [Abstract (metrics)] The abstract reports qualitative successes ('accurately recovering,' 'avoiding false positives,' 'near-zero abundances') without any quantitative error metrics, confidence intervals, or calibration checks. There are no reported numbers for bias, scatter, R², mean absolute error, or the fraction of test cases where the true value falls within the Monte Carlo Dropout confidence interval. The credibility curves are mentioned but not shown or summarized with quantitative coverage statistics. Without these, the reader cannot judge whether the stated 'accurate recovery' is meaningful, especially for parameters like surface pressure and temperature where spectral sensitivity is often degenerate. The authors should provide per-parameter and per-epoch tables of retrieval metrics, as well as calibration plots for the uncertainty estimates.
  3. [Full text (readability)] The main body of the manuscript as provided is severely corrupted: the text is a sequence of mojibake characters with no recoverable equations, figures, or section structure. This makes it impossible to verify the network architecture, the loss function, the prior ranges for molecular abundances and physical parameters, the noise injection model, the training/test split, or the derivation of the credibility curves and detectability distances. I cannot confirm that the implementation details are sound or that the reported claims follow from the described methods. The authors must resubmit a readable and complete manuscript; as it stands, the paper is not technically reviewable beyond the abstract.
  4. [Abstract (false positives)] The statement that the network 'avoids false positives and outputs near-zero abundances in scenarios of true absence' is not a well-defined claim without a decision threshold and a confusion matrix. 'Near-zero' is not a quantitative criterion; a posterior distribution concentrated at a small positive value could still be consistent with a true zero, and the fraction of test cases that would produce a false biosignature detection depends on the chosen threshold. The authors should define a detection threshold (e.g., a lower bound of the credible interval or a posterior probability cutoff) and report the false-positive rate and true-positive rate for each molecular species and epoch. Without this, the biosignature false-positive guarantee cannot be assessed.
minor comments (3)
  1. [Abstract] The phrase 'Fraunhofer A band' should be clarified: the O2 absorption near 760 nm is commonly called the Fraunhofer A band, while the 687 nm feature is the B band; the text should specify which feature is used and confirm the wavelength assignment.
  2. [Abstract] The instrument abbreviation 'HabEx/SS' is not defined in the abstract; it should be spelled out as the HabEx starshade or coronagraph configuration.
  3. [Abstract] The claim of 'operational cadence' is vague; the paper should state the actual inference time per spectrum and the expected number of targets HWO might observe, to make the computational advantage concrete.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity: the CNN is trained on synthetic spectra with known labels and tested on held-out spectra from the same forward-model family; this is supervised learning, not a self-referential derivation.

full rationale

The paper's derivation chain is a standard supervised machine-learning pipeline: synthetic, noise-injected spectra are generated from a forward model with known labels (six molecular abundances, radius, gravity, surface pressure, temperature), a 1D CNN is trained on one million such spectra, and performance is measured on unseen test spectra drawn from the same generative process. Inference via Monte Carlo Dropout and the subsequent credibility curves and detectability distances are outputs of the trained network, not inputs to its training objective. No parameter is fitted to a subset of the reported quantities and then renamed a prediction; the held-out test metrics are genuine out-of-sample evaluations within the modeled distribution. The abstract's 'mission-ready' claim does depend on an unverified assumption that the synthetic forward model and noise model are faithful to future HWO observations, but that is a generalization and robustness concern, not circularity. No load-bearing self-citation, imported uniqueness theorem, or ansatz-smuggling-through-citation is present in the evidence supplied. The paper does not define its outputs in terms of its inputs or reduce any predicted abundance to a fitted value by construction. Therefore the circularity score is 0.

Assumptions & free parameters 3 free parameters · 4 assumptions · 0 invented entities

The central claim rests on the fidelity of synthetic training data, chosen prior ranges, and instrument noise models. These are not fitted constants in the traditional sense, but they are user-chosen inputs that determine what the CNN can retrieve and how far the detectability claims extend.

free parameters (3)
  • Prior ranges for molecular abundances and physical parameters
    The simulator must sample abundances of O2, O3, CH4, CO2, H2O, and other gases across Archean, Proterozoic, and Modern Earth, plus radius, gravity, surface pressure, and temperature. These chosen ranges define the training distribution and directly bound what the CNN can retrieve.
  • Noise injection level and noise model
    The abstract says spectra are noise-injected but not at which SNR or with which noise model. This determines whether claimed detectability distances are realistic.
  • Training set class balance across the three Earth epochs
    Imbalances in how many Archean, Proterozoic, and Modern spectra were generated affect the CNN's prior and its reported performance per epoch.
assumptions (4)
  • domain assumption The synthetic forward model produces reflected-light spectra that are sufficiently accurate proxies for real HWO/LUVOIR/HabEx observations.
    All training and test data are synthetic; the entire retrieval claim inherits the forward model's fidelity. Invoked in the abstract's 'trained on over one million synthetic, noise-injected spectra'.
  • domain assumption The instrument models for LUVOIR-B (0.2-2.0 microns) and HabEx/SS (0.2-1.8 microns) capture the noise, throughput, and resolution of HWO-like observations.
    Detectability distances and credibility curves assume these instrument responses.
  • domain assumption Monte Carlo Dropout gives calibrated uncertainties.
    The abstract reports uncertainty estimation but no calibration check is mentioned in the abstract; dropout uncertainty is often miscalibrated.
  • domain assumption Integrated Gradients attributions reflect physical absorption features such as the Fraunhofer A and Hartley-Huggins bands.
    Attribution maps are used as interpretability evidence, but the abstract does not specify whether attributions were validated against known line lists or independent models.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Towards the Habitable Worlds Observatory: 1D CNN Retrieval of Reflection Spectra from Evolving Earth Analogs." pith.science (2026). https://pith.science/paper/EZQV3F5U

@misc{pith2026250800076,
  author       = {Pith},
  title        = {Pith review of: Towards the Habitable Worlds Observatory: 1D CNN Retrieval of Reflection Spectra from Evolving Earth Analogs},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/EZQV3F5U}},
  note         = {Machine review of arXiv:2508.00076}
}
abstract

Upcoming direct-imaging missions like the Habitable Worlds Observatory (HWO) aim to characterize dozens of Earth-like exoplanets by capturing their reflected-light spectra. However, traditional atmospheric retrieval frameworks are too computationally intensive to explore the high-dimensional parameter spaces such missions will generate. Here, we present a one-dimensional convolutional neural network (1D CNN), trained on over one million synthetic, noise-injected spectra simulating Archean, Proterozoic, and Modern Earth analogs, as observed by LUVOIR-B (0.2-2.0 $\mu$m) and HabEx/SS (0.2-1.8 $\mu$m). Our model simultaneously infers six molecular abundances (including biosignatures O$_2$ and O$_3$) along with radius, gravity, surface pressure, and temperature. Inference on unseen test data is performed via Monte Carlo Dropout, enabling uncertainty estimation across thousands of realizations within seconds. The network performs best where spectral features are prominent, accurately recovering CH$_4$ and CO$_2$ in Archean atmospheres and O$_2$ and O$_3$ in Modern cases, while avoiding false positives and outputting near-zero abundances in scenarios of true absence such as Archean O$_2$ and O$_3$. Interpretation via Integrated Gradients confirms that the model bases its predictions on physically meaningful features, including the Fraunhofer A band for O$_2$, and the Hartley-Huggins band for O$_3$. Credibility curve analysis indicates that O$_3$ remains retrievable across a wide range of stellar types and distances, while O$_2$ is detectable out to 12 pc around FG stars. These results elevate the CNN from proof of concept to a mission-ready retrieval engine, capable of processing direct-imaging spectra with HWO on an operational cadence.

Discussion (0). Sign in to comment.

Forward citations

Cited by 1 Pith paper

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score.

  1. An Agnostic Machine Learning Model of Photosynthetic Habitability

    astro-ph.EP 2026-06 unverdicted novelty 6.0 of 10

    An agnostic model using genetic algorithm optimization predicts photosynthetic viability declines linearly with orbital distance, expanding the photosynthetic habitable zone for M, K, and G stars beyond Earth-centric ...

Reference graph

Works this paper leans on

100 extracted references · 16 canonical work pages · cited by 1 Pith paper

  1. [1]

    write newline

    " write newline "" before.all 'output.state := FUNCTION fin.entry write newline FUNCTION new.block output.state before.all = 'skip after.block 'output.state := if FUNCTION new.sentence output.state after.block = 'skip output.state before.all = 'skip after.sentence 'output.state := if if FUNCTION not #0 #1 if FUNCTION and 'skip pop #0 if FUNCTION or pop #1...

  2. [2]

    Abadi M., et al., 2015, TensorFlow : Large-Scale Machine Learning on Heterogeneous Systems, https://www.tensorflow.org/

    2015

  3. [3]

    I., 2022, @doi [ ] 10.1051/0004-6361/202142976 , https://ui.adsabs.harvard.edu/abs/2022A&A...662A.108A 662, A108

    Ard \'e vol Mart \' nez F., Min M., Kamp I., Palmer P. I., 2022, @doi [ ] 10.1051/0004-6361/202142976 , https://ui.adsabs.harvard.edu/abs/2022A&A...662A.108A 662, A108

    doi:10.1051/0004-6361/202142976 2022

  4. [4]

    Arney G., et al., 2016, @doi [Astrobiology] 10.1089/ast.2015.1422 , https://ui.adsabs.harvard.edu/abs/2016AsBio..16..873A 16, 873

    arXiv 2016

  5. [5]

    arXiv:2309.09337

    Aubin M., et al., 2023, @doi [arXiv e-prints] 10.48550/arXiv.2309.09337 , https://ui.adsabs.harvard.edu/abs/2023arXiv230909337A p. arXiv:2309.09337

    arXiv doi:10.48550/arxiv.2309.09337 2023

  6. [6]

    Barclay T., et al., 2013, @doi [ ] 10.1038/nature11914 , https://ui.adsabs.harvard.edu/abs/2013Natur.494..452B 494, 452

    doi:10.1038/nature11914 2013

  7. [7]

    M., 2006, Pattern Recognition and Machine Learning (Information Science and Statistics)

    Bishop C. M., 2006, Pattern Recognition and Machine Learning (Information Science and Statistics). Springer-Verlag, Berlin, Heidelberg

    2006

  8. [8]

    Brack A., 1993, @doi [Origins of Life and Evolution of the Biosphere] 10.1007/BF01581985 , https://ui.adsabs.harvard.edu/abs/1993OLEB...23....3B 23, 3

    work page doi:10.1007/bf01581985 1993

Show all 100 references
  1. [9]

    Brogi M., Snellen I. A. G., de Kok R. J., Albrecht S., Birkby J., de Mooij E. J. W., 2012, @doi [ ] 10.1038/nature11161 , https://ui.adsabs.harvard.edu/abs/2012Natur.486..502B 486, 502

    2012 doi

  2. [10]

    C., Kasting J

    Catling D. C., Kasting J. F., 2017, Atmospheric Evolution on Inhabited and Lifeless Worlds . Cambridge University Press

    2017

  3. [11]

    C., Zahnle K

    Catling D. C., Zahnle K. J., 2020, @doi [Science Advances] 10.1126/sciadv.aax1420 , https://ui.adsabs.harvard.edu/abs/2020SciA....6.1420C 6, eaax1420

    2020 doi

  4. [12]

    H., 2023, @doi [RAS Techniques and Instruments] 10.1093/rasti/rzad001 , https://ui.adsabs.harvard.edu/abs/2023RASTI...2...45C 2, 45

    Changeat Q., Yip K. H., 2023, @doi [RAS Techniques and Instruments] 10.1093/rasti/rzad001 , https://ui.adsabs.harvard.edu/abs/2023RASTI...2...45C 2, 45

    2023 doi

  5. [13]

    M., Noyes R

    Charbonneau D., Brown T. M., Noyes R. W., Gilliland R. L., 2002, @doi [ ] 10.1086/338770 , https://ui.adsabs.harvard.edu/abs/2002ApJ...568..377C 568, 377

    2002 doi

  6. [14]

    Charbonneau D., et al., 2005, @doi [ ] 10.1086/429991 , https://ui.adsabs.harvard.edu/abs/2005ApJ...626..523C 626, 523

    2005 doi

  7. [15]

    Charbonneau D., et al., 2009, @doi [ ] 10.1038/nature08679 , https://ui.adsabs.harvard.edu/abs/2009Natur.462..891C 462, 891

    2009 doi

  8. [16]

    H., et al., 2021, @doi [ ] 10.3847/1538-3881/abdb36 , https://ui.adsabs.harvard.edu/abs/2021AJ....161..150C 161, 150

    Checlair J. H., et al., 2021, @doi [ ] 10.3847/1538-3881/abdb36 , https://ui.adsabs.harvard.edu/abs/2021AJ....161..150C 161, 150

    2021 doi

  9. [17]

    Chen J., Kipping D., 2017, @doi [ ] 10.3847/1538-4357/834/1/17 , https://ui.adsabs.harvard.edu/abs/2017ApJ...834...17C 834, 17

    2017 doi

  10. [18]

    Choi K., Yi J., Park C., Yoon S., 2021, @doi [IEEE Access] 10.1109/ACCESS.2021.3107975 , https://ui.adsabs.harvard.edu/abs/2021IEEEA...9l0043C 9, 120043

    2021

  11. [19]

    S., et al., 2016, @doi [Astrobiology] 10.1089/ast.2015.1295 , https://ui.adsabs.harvard.edu/abs/2016AsBio..16...89C 16, 89

    Cockell C. S., et al., 2016, @doi [Astrobiology] 10.1089/ast.2015.1295 , https://ui.adsabs.harvard.edu/abs/2016AsBio..16...89C 16, 89

    2016

  12. [20]

    Conde A., Domínguez J., 2018, @doi [Journal of Asia-Pacific Biodiversity] https://doi.org/10.1016/j.japb.2018.01.001 , 11, 161

    2018 doi

  13. [21]

    Coston C., Mamajek E., Gilbert E., 2024, in American Astronomical Society Meeting Abstracts. p. 259.15

    2024

  14. [22]

    F., 2016, @doi [ ] 10.3847/0004-637X/827/1/79 , https://ui.adsabs.harvard.edu/abs/2016ApJ...827...79C 827, 79

    Cuntz M., Guinan E. F., 2016, @doi [ ] 10.3847/0004-637X/827/1/79 , https://ui.adsabs.harvard.edu/abs/2016ApJ...827...79C 827, 79

    2016 doi

  15. [23]

    H., Meadows V

    Currie M. H., Meadows V. S., Rasmussen K. C., 2023, @doi [Planetary Science Journal] 10.3847/PSJ/accf86 , https://ui.adsabs.harvard.edu/abs/2023PSJ.....4...83C 4, 83

    2023 doi

  16. [24]

    Damiano M., Hu R., 2022, @doi [ ] 10.3847/1538-3881/ac6b97 , https://ui.adsabs.harvard.edu/abs/2022AJ....163..299D 163, 299

    2022 doi

  17. [25]

    Damiano M., Hu R., Mennesson B., 2023, @doi [ ] 10.3847/1538-3881/acefd3 , https://ui.adsabs.harvard.edu/abs/2023AJ....166..157D 166, 157

    2023 doi

  18. [26]

    I., Johnson J

    Dawson R. I., Johnson J. A., 2018, @doi [ ] 10.1146/annurev-astro-081817-051853 , https://ui.adsabs.harvard.edu/abs/2018ARA&A..56..175D 56, 175

    2018 doi

  19. [27]

    D., et al., 2019, @doi [ ] 10.3847/1538-4357/ab3be8 , https://ui.adsabs.harvard.edu/abs/2019ApJ...884...75D 884, 75

    Del Genio A. D., et al., 2019, @doi [ ] 10.3847/1538-4357/ab3be8 , https://ui.adsabs.harvard.edu/abs/2019ApJ...884...75D 884, 75

    2019 doi

  20. [28]

    D., Seager S., 2017, @doi [Journal of Geophysical Research (Planets)] 10.1002/2016JE005155 , https://ui.adsabs.harvard.edu/abs/2017JGRE..122...53D 122, 53

    Deming L. D., Seager S., 2017, @doi [Journal of Geophysical Research (Planets)] 10.1002/2016JE005155 , https://ui.adsabs.harvard.edu/abs/2017JGRE..122...53D 122, 53

    2017 doi

  21. [29]

    R., Roudier G

    Estrela R., Swain M. R., Roudier G. M., 2022, @doi [ ] 10.3847/2041-8213/aca2aa , https://ui.adsabs.harvard.edu/abs/2022ApJ...941L...5E 941, L5

    2022 doi

  22. [30]

    J., Dawson R

    Fortney J. J., Dawson R. I., Komacek T. D., 2021, @doi [Journal of Geophysical Research (Planets)] 10.1029/2020JE006629 , https://ui.adsabs.harvard.edu/abs/2021JGRE..12606629F 126, e06629

    2021 doi

  23. [31]

    arXiv:1506.02142

    Gal Y., Ghahramani Z., 2015, @doi [arXiv e-prints] 10.48550/arXiv.1506.02142 , https://ui.adsabs.harvard.edu/abs/2015arXiv150602142G p. arXiv:1506.02142

    2015 doi arXiv

  24. [32]

    S., et al., 2020, @doi [arXiv e-prints] 10.48550/arXiv.2001.06683 , https://ui.adsabs.harvard.edu/abs/2020arXiv200106683G p

    Gaudi B. S., et al., 2020, @doi [arXiv e-prints] 10.48550/arXiv.2001.06683 , https://ui.adsabs.harvard.edu/abs/2020arXiv200106683G p. arXiv:2001.06683

    2020 doi arXiv

  25. [33]

    Gelaro R., et al., 2017, @doi [Journal of Climate] 10.1175/JCLI-D-16-0758.1 , https://ui.adsabs.harvard.edu/abs/2017JCli...30.5419G 30, 5419

    2017 doi

  26. [34]

    O'Reilly Media, Incorporated

    G \'e ron A., 2022, Hands-on machine learning with Scikit-Learn, Keras, and TensorFlow: Concepts, tools, and techniques to build intelligent systems. O'Reilly Media, Incorporated

    2022

  27. [35]

    P., et al., 2010, @doi [ ] 10.1111/j.1745-3933.2010.00847.x , https://ui.adsabs.harvard.edu/abs/2010MNRAS.404L.114G 404, L114

    Gibson N. P., et al., 2010, @doi [ ] 10.1111/j.1745-3933.2010.00847.x , https://ui.adsabs.harvard.edu/abs/2010MNRAS.404L.114G 404, L114

    2010

  28. [36]

    M., et al., 2018, @doi [ ] 10.1093/mnras/stx3015 , https://ui.adsabs.harvard.edu/abs/2018MNRAS.474.5158G 474, 5158

    Goyal J. M., et al., 2018, @doi [ ] 10.1093/mnras/stx3015 , https://ui.adsabs.harvard.edu/abs/2018MNRAS.474.5158G 474, 5158

    2018 doi

  29. [37]

    Haghighipour N., 2013, @doi [Annual Review of Earth and Planetary Sciences] 10.1146/annurev-earth-042711-105340 , https://ui.adsabs.harvard.edu/abs/2013AREPS..41..469H 41, 469

    2013 doi

  30. [38]

    P., Rauer H., 2015, @doi [ ] 10.1088/2041-8205/810/2/L25 , https://ui.adsabs.harvard.edu/abs/2015ApJ...810L..25H 810, L25

    Hatzes A. P., Rauer H., 2015, @doi [ ] 10.1088/2041-8205/810/2/L25 , https://ui.adsabs.harvard.edu/abs/2015ApJ...810L..25H 810, L25

    2015 doi

  31. [39]

    W., et al., 2013, @doi [ ] 10.1038/nature12767 , https://ui.adsabs.harvard.edu/abs/2013Natur.503..381H 503, 381

    Howard A. W., et al., 2013, @doi [ ] 10.1038/nature12767 , https://ui.adsabs.harvard.edu/abs/2013Natur.503..381H 503, 381

    2013 doi

  32. [40]

    JWST Transiting Exoplanet Community Early Release Science Team et al., 2023, @doi [ ] 10.1038/s41586-022-05269-w , https://ui.adsabs.harvard.edu/abs/2023Natur.614..649J 614, 649

    2023 doi

  33. [41]

    Kaltenegger L., 2017, @doi [ ] 10.1146/annurev-astro-082214-122238 , https://ui.adsabs.harvard.edu/abs/2017ARA&A..55..433K 55, 433

    2017 doi

  34. [42]

    R., 2021, Planetary Habitability

    Kane S. R., 2021, Planetary Habitability . IOP Publishing, @doi 10.1088/2514-3433/ac2aa1

    2021 doi

  35. [43]

    F., Whitmire D

    Kasting J. F., Whitmire D. P., Reynolds R. T., 1993, @doi [ ] 10.1006/icar.1993.1010 , https://ui.adsabs.harvard.edu/abs/1993Icar..101..108K 101, 108

    1993

  36. [44]

    Kawashima Y., Rugheimer S., 2019, @doi [ ] 10.3847/1538-3881/ab14e3 , https://ui.adsabs.harvard.edu/abs/2019AJ....157..213K 157, 213

    2019 doi

  37. [45]

    L., Godolt M., Stracke B., Rauer H., 2018, @doi [Astrobiology] 10.1089/ast.2016.1632 , https://ui.adsabs.harvard.edu/abs/2018AsBio..18..116K 18, 116

    Keles E., Grenfell J. L., Godolt M., Stracke B., Rauer H., 2018, @doi [Astrobiology] 10.1089/ast.2016.1632 , https://ui.adsabs.harvard.edu/abs/2018AsBio..18..116K 18, 116

    2018

  38. [46]

    arXiv:2411.12708

    Kofman V., Villanueva G., Fauchez T., Mandell A., Johnson T., Payne A., Latouf N., Kelkar S., 2024, arXiv e-prints, https://ui.adsabs.harvard.edu/abs/2024arXiv241112708K p. arXiv:2411.12708

    2024 arXiv

  39. [47]

    K., et al., 2013, @doi [ ] 10.1088/0004-637X/765/2/131 , https://ui.adsabs.harvard.edu/abs/2013ApJ...765..131K 765, 131

    Kopparapu R. K., et al., 2013, @doi [ ] 10.1088/0004-637X/765/2/131 , https://ui.adsabs.harvard.edu/abs/2013ApJ...765..131K 765, 131

    2013 doi

  40. [48]

    K., Ramirez R

    Kopparapu R. K., Ramirez R. M., SchottelKotte J., Kasting J. F., Domagal-Goldman S., Eymet V., 2014, @doi [ ] 10.1088/2041-8205/787/2/L29 , https://ui.adsabs.harvard.edu/abs/2014ApJ...787L..29K 787, L29

    2014 doi

  41. [49]

    Kreidberg L., et al., 2014, @doi [ ] 10.1088/2041-8205/793/2/L27 , https://ui.adsabs.harvard.edu/abs/2014ApJ...793L..27K 793, L27

    2014 doi

  42. [50]

    Lavie B., et al., 2017, @doi [ ] 10.3847/1538-3881/aa7ed8 , https://ui.adsabs.harvard.edu/abs/2017AJ....154...91L 154, 91

    2017 doi

  43. [51]

    S., Henderson D., Howard R

    LeCun Y., Boser B., Denker J. S., Henderson D., Howard R. E., Hubbard W., Jackel L. D., 1989, @doi [Neural Computation] 10.1162/neco.1989.1.4.541 , 1, 541

    1989 doi

  44. [52]

    D., 2001, @doi [Oecologia] 10.1007/s004420100716 , 129, 271

    Legendre P., Gallagher E. D., 2001, @doi [Oecologia] 10.1007/s004420100716 , 129, 271

    2001 doi

  45. [53]

    A., 2019, @doi [IEEE Transactions on Geoscience and Remote Sensing] 10.1109/TGRS.2019.2907932 , https://ui.adsabs.harvard.edu/abs/2019ITGRS..57.6690L 57, 6690

    Li S., Song W., Fang L., Chen Y., Ghamisi P., Benediktsson J. A., 2019, @doi [IEEE Transactions on Geoscience and Remote Sensing] 10.1109/TGRS.2019.2907932 , https://ui.adsabs.harvard.edu/abs/2019ITGRS..57.6690L 57, 6690

    2019

  46. [54]

    W., Reinhard C

    Lyons T. W., Reinhard C. T., Planavsky N. J., 2014, @doi [ ] 10.1038/nature13068 , https://ui.adsabs.harvard.edu/abs/2014Natur.506..307L 506, 307

    2014 doi

  47. [55]

    Madhusudhan N., Seager S., 2009, @doi [ ] 10.1088/0004-637X/707/1/24 , https://ui.adsabs.harvard.edu/abs/2009ApJ...707...24M 707, 24

    2009 doi

  48. [56]

    P., Obermeier C., 2022, @doi [ ] 10.1093/mnras/stab3692 , https://ui.adsabs.harvard.edu/abs/2022MNRAS.513.5505M 513, 5505

    Malik A., Moster B. P., Obermeier C., 2022, @doi [ ] 10.1093/mnras/stab3692 , https://ui.adsabs.harvard.edu/abs/2022MNRAS.513.5505M 513, 5505

    2022 doi

  49. [57]

    E., 2019, A Modern Mean Dwarf Stellar Colour and Effective Temperature Sequence, https://www.pas.rochester.edu/ emamajek/EEM_dwarf_UBVIJHK_colors_Teff.dat

    Mamajek E. E., 2019, A Modern Mean Dwarf Stellar Colour and Effective Temperature Sequence, https://www.pas.rochester.edu/ emamajek/EEM_dwarf_UBVIJHK_colors_Teff.dat

    2019

  50. [58]

    R., 2006, @doi [Annual Review of Earth and Planetary Sciences] 10.1146/annurev.earth.33.031504.103001 , https://ui.adsabs.harvard.edu/abs/2006AREPS..34..355M 34, 355

    Marshall C. R., 2006, @doi [Annual Review of Earth and Planetary Sciences] 10.1146/annurev.earth.33.031504.103001 , https://ui.adsabs.harvard.edu/abs/2006AREPS..34..355M 34, 355

    2006

  51. [59]

    Mayor M., Queloz D., 1995, @doi [ ] 10.1038/378355a0 , https://ui.adsabs.harvard.edu/abs/1995Natur.378..355M 378, 355

    1995 doi

  52. [60]

    McIntyre S. R. N., King P. L., Mills F. P., 2023, @doi [ ] 10.1093/mnras/stad095 , https://ui.adsabs.harvard.edu/abs/2023MNRAS.519.6210M 519, 6210

    2023 doi

  53. [61]

    N., 2016, @doi [Journal of Geophysical Research (Planets)] 10.1002/2016JE005088 , https://ui.adsabs.harvard.edu/abs/2016JGRE..121.1962M 121, 1962

    Morbidelli A., Raymond S. N., 2016, @doi [Journal of Geophysical Research (Planets)] 10.1002/2016JE005088 , https://ui.adsabs.harvard.edu/abs/2016JGRE..121.1962M 121, 1962

    2016 doi

  54. [62]

    M \"u ller S., Baron J., Helled R., Bouchy F., Parc L., 2024, @doi [ ] 10.1051/0004-6361/202348690 , https://ui.adsabs.harvard.edu/abs/2024A&A...686A.296M 686, A296

    2024 doi

  55. [63]

    The National Academies Press , Washington, DC , @doi 10.17226/26141

    National Academies of Sciences and Medicine, Engineering 2021, Pathways to Discovery in Astronomy and Astrophysics for the 2020s . The National Academies Press , Washington, DC , @doi 10.17226/26141

    2021 doi

  56. [64]

    C., Madhusudhan N., 2020, @doi [ ] 10.1093/mnras/staa1150 , https://ui.adsabs.harvard.edu/abs/2020MNRAS.496..269N 496, 269

    Nixon M. C., Madhusudhan N., 2020, @doi [ ] 10.1093/mnras/staa1150 , https://ui.adsabs.harvard.edu/abs/2020MNRAS.496..269N 496, 269

    2020 doi

  57. [65]

    O'Malley T., Bursztein E., Long J., Chollet F., Jin H., Invernizzi L., et al., 2019, KerasTuner, https://github.com/keras-team/keras-tuner

    2019

  58. [66]

    F., Dorn C., Helled R., Bouchy F., Haldemann J., Alibert Y., 2020, @doi [ ] 10.1051/0004-6361/202038006 , https://ui.adsabs.harvard.edu/abs/2020A&A...640A.135O 640, A135

    Otegi J. F., Dorn C., Helled R., Bouchy F., Haldemann J., Alibert Y., 2020, @doi [ ] 10.1051/0004-6361/202038006 , https://ui.adsabs.harvard.edu/abs/2020A&A...640A.135O 640, A135

    2020 doi

  59. [67]

    D., ed., Astrophysics and Space Science Library Vol

    Owen T., 1980, in Papagiannis M. D., ed., Astrophysics and Space Science Library Vol. 83, Strategies for the Search for Life in the Universe. p. 177, @doi 10.1007/978-94-009-9115-6_17

    1980 doi

  60. [68]

    H., Rahulamathavan Y., Dogan S., Kondoz A

    Pathmaperuma M. H., Rahulamathavan Y., Dogan S., Kondoz A. M., 2022, @doi [Sensors] 10.3390/s22197643 , https://ui.adsabs.harvard.edu/abs/2022Senso..22.7643P 22, 7643

    2022 doi

  61. [69]

    Pedregosa F., et al., 2011, @doi [Journal of Machine Learning Research] 10.48550/arXiv.1201.0490 , https://ui.adsabs.harvard.edu/abs/2011JMLR...12.2825P 12, 2825

    2011 doi arXiv

  62. [70]

    W., et al., 2020, @doi [ ] 10.1051/0004-6361/201937381 , https://ui.adsabs.harvard.edu/abs/2020A&A...637A..38P 637, A38

    Phillips M. W., et al., 2020, @doi [ ] 10.1051/0004-6361/201937381 , https://ui.adsabs.harvard.edu/abs/2020A&A...637A..38P 637, A38

    2020 doi

  63. [71]

    J., et al., 2014, @doi [Nature Geoscience] 10.1038/ngeo2122 , https://ui.adsabs.harvard.edu/abs/2014NatGe...7..283P 7, 283

    Planavsky N. J., et al., 2014, @doi [Nature Geoscience] 10.1038/ngeo2122 , https://ui.adsabs.harvard.edu/abs/2014NatGe...7..283P 7, 283

    2014 doi

  64. [72]

    D., Salvador A., 2023, @doi [Planetary Science Journal] 10.3847/PSJ/acac9a , https://ui.adsabs.harvard.edu/abs/2023PSJ.....4...10R 4, 10

    Robinson T. D., Salvador A., 2023, @doi [Planetary Science Journal] 10.3847/PSJ/acac9a , https://ui.adsabs.harvard.edu/abs/2023PSJ.....4...10R 4, 10

    2023 doi

  65. [73]

    D., Fortney J

    Robinson T. D., Fortney J. J., Hubbard W. B., 2017, @doi [ ] 10.3847/1538-4357/aa951e , https://ui.adsabs.harvard.edu/abs/2017ApJ...850..128R 850, 128

    2017 doi

  66. [74]

    Rugheimer S., Kaltenegger L., 2018, @doi [ ] 10.3847/1538-4357/aaa47a , https://ui.adsabs.harvard.edu/abs/2018ApJ...854...19R 854, 19

    2018 doi

  67. [75]

    E., Hinton G

    Rumelhart D. E., Hinton G. E., Williams R. J., 1985, Technical report, Learning Internal Representations by Error Propagation. Institute for Cognitive Science, University of California, San Diego

    1985

  68. [76]

    W., et al., 2018, @doi [Astrobiology] 10.1089/ast.2017.1729 , https://ui.adsabs.harvard.edu/abs/2018AsBio..18..663S 18, 663

    Schwieterman E. W., et al., 2018, @doi [Astrobiology] 10.1089/ast.2017.1729 , https://ui.adsabs.harvard.edu/abs/2018AsBio..18..663S 18, 663

    2018

  69. [77]

    Princeton University Press

    Seager S., 2010, Exoplanet Atmospheres: Physical Processes . Princeton University Press

    2010

  70. [78]

    Seager S., 2013, @doi [Science] 10.1126/science.1232226 , https://ui.adsabs.harvard.edu/abs/2013Sci...340..577S 340, 577

    2013 doi

  71. [79]

    J., Vanderburg A., 2018, @doi [ ] 10.3847/1538-3881/aa9e09 , https://ui.adsabs.harvard.edu/abs/2018AJ....155...94S 155, 94

    Shallue C. J., Vanderburg A., 2018, @doi [ ] 10.3847/1538-3881/aa9e09 , https://ui.adsabs.harvard.edu/abs/2018AJ....155...94S 155, 94

    2018 doi

  72. [80]

    arXiv:1703.01365

    Sundararajan M., Taly A., Yan Q., 2017, @doi [arXiv e-prints] 10.48550/arXiv.1703.01365 , https://ui.adsabs.harvard.edu/abs/2017arXiv170301365S p. arXiv:1703.01365

    2017 doi arXiv

  73. [81]

    arXiv:1912.06219

    The LUVOIR Team 2019, @doi [arXiv e-prints] 10.48550/arXiv.1912.06219 , https://ui.adsabs.harvard.edu/abs/2019arXiv191206219T p. arXiv:1912.06219

    2019 doi arXiv

  74. [82]

    arXiv:1411.4280

    Tompson J., Goroshin R., Jain A., LeCun Y., Bregler C., 2014, @doi [arXiv e-prints] 10.48550/arXiv.1411.4280 , https://ui.adsabs.harvard.edu/abs/2014arXiv1411.4280T p. arXiv:1411.4280

    2014 doi arXiv

  75. [83]

    P., Tinetti G., Tennyson J., Yurchenko S

    Tsiaras A., Waldmann I. P., Tinetti G., Tennyson J., Yurchenko S. N., 2019, @doi [Nature Astronomy] 10.1038/s41550-019-0878-9 , https://ui.adsabs.harvard.edu/abs/2019NatAs...3.1086T 3, 1086

    2019 doi

  76. [84]

    Valizadegan H., et al., 2022, @doi [ ] 10.3847/1538-4357/ac4399 , https://ui.adsabs.harvard.edu/abs/2022ApJ...926..120V 926, 120

    2022 doi

  77. [85]

    L., Smith M

    Villanueva G. L., Smith M. D., Protopapa S., Faggi S., Mandell A. M., 2018, @doi [ ] 10.1016/j.jqsrt.2018.05.023 , https://ui.adsabs.harvard.edu/abs/2018JQSRT.217...86V 217, 86

    2018 doi

  78. [86]

    L., Liuzzi G., Faggi S., Protopapa S., Kofman V., Fauchez T., Stone S

    Villanueva G. L., Liuzzi G., Faggi S., Protopapa S., Kofman V., Fauchez T., Stone S. W., Mandell A. M., 2022, Fundamentals of the Planetary Spectrum Generator . NASA Goddard Space Flight Center

    2022

  79. [87]

    P., 2016, @doi [ ] 10.3847/0004-637X/820/2/107 , https://ui.adsabs.harvard.edu/abs/2016ApJ...820..107W 820, 107

    Waldmann I. P., 2016, @doi [ ] 10.3847/0004-637X/820/2/107 , https://ui.adsabs.harvard.edu/abs/2016ApJ...820..107W 820, 107

    2016 doi

  80. [88]

    P., Tinetti G., Rocchetto M., Barton E

    Waldmann I. P., Tinetti G., Rocchetto M., Barton E. J., Yurchenko S. N., Tennyson J., 2015, @doi [ ] 10.1088/0004-637X/802/2/107 , https://ui.adsabs.harvard.edu/abs/2015ApJ...802..107W 802, 107

    2015 doi

  81. [89]

    J., et al., 2017, @doi [ ] 10.3847/1538-4365/aa7a06 , https://ui.adsabs.harvard.edu/abs/2017ApJS..231...12W 231, 12

    Way M. J., et al., 2017, @doi [ ] 10.3847/1538-4365/aa7a06 , https://ui.adsabs.harvard.edu/abs/2017ApJS..231...12W 231, 12

    2017 doi

  82. [90]

    K., Kumar M., Jain A., Morato J., 2024, @doi [Earth Science Informatics] 10.1007/s12145-024-01301-x , https://ui.adsabs.harvard.edu/abs/2024EScIn.tmp..106Y

    Yadav D., Kapoor K., Yadav A. K., Kumar M., Jain A., Morato J., 2024, @doi [Earth Science Informatics] 10.1007/s12145-024-01301-x , https://ui.adsabs.harvard.edu/abs/2024EScIn.tmp..106Y

    2024 doi

  83. [91]

    H., Changeat Q., Nikolaou N., Morvan M., Edwards B., Waldmann I

    Yip K. H., Changeat Q., Nikolaou N., Morvan M., Edwards B., Waldmann I. P., Tinetti G., 2021, @doi [ ] 10.3847/1538-3881/ac1744 , 162, 195

    2021 doi

  84. [92]

    H., Changeat Q., Al-Refaie A., Waldmann I

    Yip K. H., Changeat Q., Al-Refaie A., Waldmann I. P., 2024, @doi [ ] 10.3847/1538-4357/ad063f , https://ui.adsabs.harvard.edu/abs/2024ApJ...961...30Y 961, 30

    2024 doi

  85. [93]

    E., Shull J

    Zahnle K., 1998, in Woodward C. E., Shull J. M., Thronson Harley A. J., eds, Astronomical Society of the Pacific Conference Series Vol. 148, Origins. p. 364

    1998

  86. [94]

    J., Catling D

    Zahnle K. J., Catling D. C., 2013, in 44th Annual Lunar and Planetary Science Conference. Lunar and Planetary Science Conference. p. 2787

    2013

  87. [95]

    J., Catling D

    Zahnle K. J., Catling D. C., 2017, @doi [ ] 10.3847/1538-4357/aa7846 , https://ui.adsabs.harvard.edu/abs/2017ApJ...843..122Z 843, 122

    2017 doi

  88. [96]

    I., Pan S., Aggarwal C

    Zamanzadeh Darban Z., Webb G. I., Pan S., Aggarwal C. C., Salehi M., 2022, @doi [arXiv e-prints] 10.48550/arXiv.2211.05244 , https://ui.adsabs.harvard.edu/abs/2022arXiv221105244Z p. arXiv:2211.05244

    2022 doi arXiv

  89. [97]

    arXiv:1706.09092

    Zhang J., Xia Y., Wu Q., Xie Y., 2017, @doi [arXiv e-prints] 10.48550/arXiv.1706.09092 , https://ui.adsabs.harvard.edu/abs/2017arXiv170609092Z p. arXiv:1706.09092

    2017 doi arXiv

  90. [98]

    Zhang M., Chachan Y., Kempton E. M. R., Knutson H. A., 2019, @doi [ ] 10.1088/1538-3873/aaf5ad , https://ui.adsabs.harvard.edu/abs/2019PASP..131c4501Z 131, 034501

    2019 doi

  91. [99]

    P., 2018, @doi [ ] 10.3847/1538-3881/aae77c , https://ui.adsabs.harvard.edu/abs/2018AJ....156..268Z 156, 268

    Zingales T., Waldmann I. P., 2018, @doi [ ] 10.3847/1538-3881/aae77c , https://ui.adsabs.harvard.edu/abs/2018AJ....156..268Z 156, 268

    2018 doi

  92. [100]

    Zorzan S., et al., 2025, @doi [The Astrophysical Journal Supplement Series] 10.3847/1538-4365/adb03a , 277, 38

    2025 doi

Pith tools

Reviewed August 6, 2026 · model on record in the stance chip above.