pith. sign in

arxiv: 2605.31420 · v1 · pith:V64UZQRDnew · submitted 2026-05-29 · 🌌 astro-ph.GA

Enhancing Galaxy Classification with U-Net Variational Autoencoders. III. Disk-like Galaxy Identification in JWST Samples of up to redshift 4

Sergey S. Mirzoyan , Arno Avagyan This is my paper

Pith reviewed 2026-06-28 22:00 UTC · model grok-4.3

classification 🌌 astro-ph.GA
keywords JWSTgalaxy morphologyhigh-redshift galaxiesdisk galaxiesmachine learningU-Netvariational autoencodergalaxy classification
0
0 comments X

The pith

Disk-like galaxies are identified among massive systems in JWST data up to redshift 4 using a U-Net Variational Autoencoder.

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

This paper applies a previously developed U-Net Variational Autoencoder classifier to a sample of 1,380 massive galaxies with stellar masses above 10^10 solar masses observed by JWST across redshifts 0.5 to 4. The classifier identifies 382 disk-like galaxies in this population. A sympathetic reader would care because the detections indicate that such structures exist across a wide redshift range where they have traditionally been considered rare. The work illustrates how denoising combined with machine learning can extract morphological information from large high-redshift imaging surveys.

Core claim

Within this population our classifier detects 382 disk-like galaxies with a subset showing uncertain features. This expanded dataset allows examination of the distribution of disk-like systems in a statistically meaningful high-redshift regime including epochs where well-ordered disks are traditionally expected to be rare. The results demonstrate that disk-like structures persist across a broad range of redshifts and stellar masses suggesting that massive disks may be more common in the early universe than previously assumed.

What carries the argument

The U-Net Variational Autoencoder classifier applied after denoising to detect disk-like morphologies in JWST images.

If this is right

  • Disk-like galaxies exist in statistically meaningful numbers at redshifts where they were expected to be rare.
  • Massive disks appear across a broad range of stellar masses and redshifts up to 4.
  • Denoising plus machine-learning classification enables morphological studies of large JWST galaxy samples.
  • Traditional expectations about the rarity of ordered disks at early times may need revision.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • The detections could imply that disk assembly processes operated efficiently at earlier cosmic times than many formation models currently allow.
  • Applying the same pipeline to even higher-redshift or lower-mass samples might map the onset of disk formation more precisely.
  • Cross-checking the machine-learning labels against independent morphological indicators such as kinematic data would strengthen or refute the morphological assignments.

Load-bearing premise

The U-Net Variational Autoencoder classifier identifies disk-like morphologies in high-redshift JWST images without major biases from noise resolution or redshift-dependent appearance changes.

What would settle it

A systematic visual or spectroscopic reclassification of the 382 candidates that finds most are not disks due to misidentification from redshift effects or noise would falsify the persistence claim.

Figures

Figures reproduced from arXiv: 2605.31420 by Arno Avagyan, Sergey S. Mirzoyan.

Figure 1
Figure 1. Figure 1: Redshift distribution of the 1,380 galaxy sample shown as a histogram, with redshift values plotted along the x-axis and galaxy counts per bin along the y-axis. In addition to the JWST/NIRCam imaging from the CEERS survey described above, we employ two auxiliary datasets to train and validate the components of our denoising and morphology–classification [PITH_FULL_IMAGE:figures/full_fig_p004_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: Representative examples of JWST cutout images rejected by the automated quality-assessment [PITH_FULL_IMAGE:figures/full_fig_p005_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: Image rejection rates from the quality-assessment pipeline. [PITH_FULL_IMAGE:figures/full_fig_p006_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: Visual validation of the denoising and classification pipeline applied to JWST galaxy images. The [PITH_FULL_IMAGE:figures/full_fig_p008_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: Disk fraction versus redshift bin with Wilson 95% confidence intervals. [PITH_FULL_IMAGE:figures/full_fig_p009_5.png] view at source ↗
Figure 6
Figure 6. Figure 6: Disk fraction versus stellar mass bin with Wilson 95% confidence intervals. [PITH_FULL_IMAGE:figures/full_fig_p010_6.png] view at source ↗
read the original abstract

In this third study of the series, we extend our U-Net Variational Autoencoder-based galaxy classification framework to a significantly larger JWST sample spanning the redshift range $0.5 < z < 4$. Focusing on massive systems with stellar masses exceeding $10^{10}\,M_\odot$, we analyze 1,380 galaxies that satisfy these criteria and apply our previously developed denoising and classification pipeline to identify disk-like morphologies across cosmic time. Within this population, our classifier detects 382 disk-like galaxies, with a subset showing uncertain features consistent with the expected performance limits of current deep-learning models. This expanded dataset allows us to examine the distribution of disk-like systems in a statistically meaningful high-redshift regime, including epochs where well-ordered disks are traditionally expected to be rare. The results demonstrate that disk-like structures persist across a broad range of redshifts and stellar masses, suggesting that massive disks may be more common in the early universe than previously assumed. These findings emphasize the value of combining advanced denoising techniques with machine-learning-based morphological analysis for characterizing galaxy populations in large JWST surveys.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

1 major / 1 minor

Summary. The paper applies a previously developed U-Net Variational Autoencoder denoising and classification pipeline to 1,380 massive galaxies (M_* > 10^10 M_⊙) in JWST data spanning 0.5 < z < 4, reports detection of 382 disk-like systems, and concludes that disk-like morphologies persist across this redshift range and are more common at high redshift than traditionally assumed.

Significance. If the classifier returns unbiased labels, the result would indicate that well-ordered massive disks form earlier than expected in standard models, with implications for the timing of disk assembly and the role of mergers versus in-situ growth. The work also illustrates a scalable ML approach for morphological classification in large JWST surveys.

major comments (1)
  1. [Abstract] Abstract: The central claim that disk-like structures persist to z=4 and are more common than assumed rests on the U-Net VAE returning accurate disk/non-disk labels on JWST images at 0.5<z<4, yet the manuscript supplies no quantitative validation (confusion matrices, false-positive rates, recovery fractions versus redshift or stellar mass, or tests against simulated high-z images with cosmological dimming and band-shifting) to demonstrate that performance limits do not introduce redshift-dependent biases.
minor comments (1)
  1. [Abstract] Abstract: The phrase 'a subset showing uncertain features consistent with the expected performance limits' is stated without accompanying numbers or a breakdown by redshift bin, which would help readers assess the fraction of the sample affected.

Simulated Author's Rebuttal

1 responses · 0 unresolved

We thank the referee for their constructive feedback. We respond to the major comment below.

read point-by-point responses

  1. Referee: [Abstract] Abstract: The central claim that disk-like structures persist to z=4 and are more common than assumed rests on the U-Net VAE returning accurate disk/non-disk labels on JWST images at 0.5<z<4, yet the manuscript supplies no quantitative validation (confusion matrices, false-positive rates, recovery fractions versus redshift or stellar mass, or tests against simulated high-z images with cosmological dimming and band-shifting) to demonstrate that performance limits do not introduce redshift-dependent biases.

    Authors: We agree that the current manuscript does not supply new quantitative validation metrics (such as confusion matrices or recovery fractions versus redshift) specifically for the JWST sample at 0.5 < z < 4. The U-Net VAE pipeline and its performance were validated in Papers I and II of the series. To address the referee's concern about potential redshift-dependent biases from cosmological dimming and band-shifting, we will add these quantitative tests using simulated high-redshift images in a revised version of the manuscript. revision: yes

Circularity Check

0 steps flagged

No circularity: observational application of prior classifier to independent JWST sample

full rationale

The paper's chain consists of applying a previously developed U-Net VAE denoising and classification pipeline (from parts I and II) to a new, independent sample of 1380 massive galaxies at 0.5 < z < 4 drawn from JWST observations, then reporting the count of 382 disk-like systems and interpreting the distribution. This is a direct application to external data rather than any derivation that reduces by construction to fitted parameters, self-defined quantities, or a self-citation chain. No equations or steps in the provided text exhibit self-definitional loops, fitted inputs renamed as predictions, or uniqueness theorems imported from the same authors. The result is therefore self-contained against the observational inputs.

Axiom & Free-Parameter Ledger

0 free parameters · 0 axioms · 0 invented entities

Abstract supplies no explicit free parameters, axioms, or invented entities; the method and its assumptions are referenced to earlier papers in the series.

pith-pipeline@v0.9.1-grok · 5736 in / 989 out tokens · 25696 ms · 2026-06-28T22:00:16.600485+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

64 extracted references · 13 canonical work pages · 3 internal anchors

  1. [1]

    L., et al

    Andalman, Z. L., et al. 2025, On the origin of the high star formation efficiency in massive galaxies at Cosmic Dawn, MNRAS, 540, 3350

    2025

  2. [2]

    T., et al

    Atemkeng, M. T., et al. 2025, A benchmark analysis of saliency-based explainable deep learning methods for the morphological classification of radio galaxies, arXiv:2502.17207

    work page arXiv 2025

  3. [3]

    2011, The EFIGI catalogue of 4458 nearby galaxies with detailed morphology, A&A, 532, A74

    Baillard, A., et al. 2011, The EFIGI catalogue of 4458 nearby galaxies with detailed morphology, A&A, 532, A74

    2011

  4. [4]

    2025, The Blue Jay Survey: Deep JWST spectroscopy for a representative sample of galaxies at cosmic noon, arXiv:2510.11775

    Belli, S., et al. 2025, The Blue Jay Survey: Deep JWST spectroscopy for a representative sample of galaxies at cosmic noon, arXiv:2510.11775

    work page arXiv 2025

  5. [5]

    F., et al

    Bell, E. F., et al. 2026, The low-mass and structured stellar halo of M83 argues against a merger origin for its starburst and extended neutral hydrogen disk, ApJ, 997, 153

    2026

  6. [6]

    Buta, R. J. 2019, The systematics of galaxy morphology in the de Vaucouleurs revised Hubble–Sandage classification system: application to the EFIGI sample, MNRAS, 488, 590

    2019

  7. [7]

    2026, JWST Advanced Deep Extragalactic Survey (JADES) Data Release 5: Catalogs of inferred morphological properties, arXiv:2601.15957

    Carreira, C., et al. 2026, JWST Advanced Deep Extragalactic Survey (JADES) Data Release 5: Catalogs of inferred morphological properties, arXiv:2601.15957

    work page arXiv 2026

  8. [8]

    L., et al

    Finkelstein, S. L., et al. 2025, The Cosmic Evolution Early Release Science Survey (CEERS), ApJL

    2025

  9. [9]

    S., & Welling, M

    Cohen, T. S., & Welling, M. 2016, Group Equivariant Convolutional Networks, in Proc. of the 33rd International Conference on Machine Learning (ICML), 2990

    2016

  10. [10]

    Conselice, C. J. 2014, The evolution of galaxy structure over cosmic time, ARA&A, 52, 291

    2014

  11. [11]

    Conselice, C. J. 2020, The Cosmic Evolution of Galaxy Structure, IOP Publishing

    2020

  12. [12]

    2023, Discovery of the highest redshift galaxies with JWST, AAS Meeting Abstracts, 241, 153–07

    Conselice, C., et al. 2023, Discovery of the highest redshift galaxies with JWST, AAS Meeting Abstracts, 241, 153–07

    2023

  13. [13]

    J., et al

    Conselice, C. J., et al. 2024, EPOCHS Paper V: The dependence of galaxy formation on galaxy structure at z<7 from JWST, MNRAS, 531, 4857

    2024

  14. [14]

    2026, An updated efficient galaxy morphology classification model based on ConvNeXt encoding with UMAP dimensionality reduction, AJ, 171, 59

    Fang, G., et al. 2026, An updated efficient galaxy morphology classification model based on ConvNeXt encoding with UMAP dimensionality reduction, AJ, 171, 59

    2026

  15. [15]

    P., et al

    Gardner, J. P., et al. 2006, The James Webb Space Telescope, Space Sci. Rev., 123, 485

    2006

  16. [16]

    2018, The size evolution of star-forming and quenched galaxies in the IllustrisTNG simulation, MNRAS, 474, 3976

    Genel, S., et al. 2018, The size evolution of star-forming and quenched galaxies in the IllustrisTNG simulation, MNRAS, 474, 3976

    2018

  17. [17]

    2025, DAWN JWST Archive: Morphology from profile fitting of over 340,000 galaxies, arXiv:2505.21622

    Genin, A., et al. 2025, DAWN JWST Archive: Morphology from profile fitting of over 340,000 galaxies, arXiv:2505.21622

    work page arXiv 2025

  18. [18]

    2017, The Early Evolution of Galactic Disks: Structure, Dynamics and Feedback, Disk Instabilities Across Cosmic Scales, 5

    Genzel, R. 2017, The Early Evolution of Galactic Disks: Structure, Dynamics and Feedback, Disk Instabilities Across Cosmic Scales, 5

    2017

  19. [19]

    2025, Galaxy Morphological Classification with Zernike Moments and Machine Learning, ApJS, 277, 10

    Ghaderi, H., et al. 2025, Galaxy Morphological Classification with Zernike Moments and Machine Learning, ApJS, 277, 10

    2025

  20. [20]

    A., et al

    Grogin, N. A., et al. 2011, The Cosmic Assembly Near-infrared Deep Extragalactic Legacy Survey (CANDELS), HST WFC3/ACS Deep Survey Description, Survey overview available at https://www.ipac.caltech.edu/project/candels

    2011

  21. [21]

    2017, On Calibration of Modern Neural Networks, in Proc

    Guo, C., et al. 2017, On Calibration of Modern Neural Networks, in Proc. of the 34th International Conference on Machine Learning (ICML), 1321

    2017

  22. [22]

    and Ozernoy L.M., 1981, Accretion of the cloud of gas debris of stars disrupted by the tidal forces of a supermassive black hole, A&A, 95, 39

    Gurzadyan, V.G. and Ozernoy L.M., 1981, Accretion of the cloud of gas debris of stars disrupted by the tidal forces of a supermassive black hole, A&A, 95, 39

    1981

  23. [23]

    Auto-Encoding Variational Bayes

    Kingma, D. P., & Welling, M. 2014, Auto-Encoding Variational Bayes, arXiv:1312.6114

    work page internal anchor Pith review Pith/arXiv arXiv 2014

  24. [24]

    2024, Hot gas accretion fuels star formation faster than cold accretion in high-redshift galaxies, MNRAS, 534, 918

    Kocjan, Z., et al. 2024, Hot gas accretion fuels star formation faster than cold accretion in high-redshift galaxies, MNRAS, 534, 918

    2024

  25. [25]

    2024, JWST reveals a surprisingly high fraction of galaxies being spiral-like at 0.5 z 4 , ApJL, 968, L15

    Kuhn, V., et al. 2024, JWST reveals a surprisingly high fraction of galaxies being spiral-like at 0.5 z 4 , ApJL, 968, L15

    2024

  26. [26]

    H., Park, C., Hwang, H

    Lee, J. H., Park, C., Hwang, H. S., & Kwon, M. 2024, Morphology of Galaxies in JWST Fields: Initial Distribution and Evolution of Galaxy Morphology, ApJ, 936, 112

    2024

  27. [27]

    2023, Cold gas disks in main-sequence galaxies at cosmic noon, A&A, 672, A106

    Lelli, F., et al. 2023, Cold gas disks in main-sequence galaxies at cosmic noon, A&A, 672, A106

    2023

  28. [28]

    2011, Galaxy Zoo 1: data release of morphological classifications for nearly 900,000 galaxies, MNRAS, 410, 166

    Lintott, C., et al. 2011, Galaxy Zoo 1: data release of morphological classifications for nearly 900,000 galaxies, MNRAS, 410, 166

    2011

  29. [29]

    2022, A ConvNet for the 2020s, in Proc

    Liu, Z., et al. 2022, A ConvNet for the 2020s, in Proc. of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 11976

    2022

  30. [30]

    M., et al

    Lotz, J. M., et al. 2004, A new nonparametric approach to galaxy morphological classification, AJ, 128, 163

    2004

  31. [31]

    A., et al

    Mager, V. A., et al. 2018, Galaxy structure in the ultraviolet: wavelength dependence of morphological parameters, ApJ, 864, 123

    2018

  32. [32]

    UMAP: Uniform Manifold Approximation and Projection for Dimension Reduction

    McInnes, L., et al. 2018, UMAP: Uniform Manifold Approximation and Projection for Dimension Reduction, arXiv:1802.03426

    work page internal anchor Pith review Pith/arXiv arXiv 2018

  33. [33]

    A., et al

    Meyer, R. A., et al. 2025, JWST COSMOS-3D: Spectroscopic census of [O III] emitters at 6.75<z<9.05 , arXiv:2510.11373

    work page arXiv 2025

  34. [34]

    S., et al

    Mirzoyan, S. S., et al. 2019, Machine learning and Kolmogorov analysis to reveal gravitational lenses, MNRAS Lett., 489, L32

    2019

  35. [35]

    Mirzoyan, S. S. 2025a, Enhancing Galaxy Classification with U-Net VAEs for Denoising, RAA, 25, 095006

  36. [36]

    Mirzoyan, S. S. 2025b, Enhancing Galaxy Classification with U-Net VAEs. II. JWST High-Redshift Sample, RAA, 26, 025010

  37. [37]

    Disk-like galaxies at 4 < z < 7.7 : JWST/NIRCam morphologies revealed by denoising VAE-GCNN classification

    Mirzoyan, S. S., Avagyan, A., & Gurzadyan, V. G. 2026, Disk-like Galaxies at 4 < z < 7.7 : JWST/NIRCam Morphologies Revealed by Denoising VAE--GCNN Classification, MNRAS Lett., in press, arXiv:2604.14599

    work page internal anchor Pith review Pith/arXiv arXiv 2026

  38. [38]

    2010, Galaxy Formation and Evolution, Cambridge Univ

    Mo, H., et al. 2010, Galaxy Formation and Evolution, Cambridge Univ. Press

    2010

  39. [39]

    Naab, T., & Ostriker, J. P. 2017, Theoretical challenges in galaxy formation, ARA&A, 55, 59

    2017

  40. [40]

    W., et al

    Nightingale, J. W., et al. 2023, PyAutoLens: Automated Modeling of Strong Gravitational Lenses, JOSS, 8, 5073

    2023

  41. [41]

    2023, E(2) -equivariant neural networks for robust galaxy morphology classification, arXiv:2311.01500

    Pandya, S., et al. 2023, E(2) -equivariant neural networks for robust galaxy morphology classification, arXiv:2311.01500

    work page arXiv 2023

  42. [42]

    1988, TTidal disruption of stars by black holes of 106-108 solar masses in nearby galaxies, Nature, 333, 523

    Rees, M. 1988, TTidal disruption of stars by black holes of 106-108 solar masses in nearby galaxies, Nature, 333, 523

    1988

  43. [43]

    J., et al

    Rieke, M. J., et al. 2005, Overview of the James Webb Space Telescope Near-Infrared Camera (NIRCam), Proc. SPIE, 5904, 1

    2005

  44. [44]

    J., et al

    Rieke, M. J., et al. 2023, The JWST Near-Infrared Camera (NIRCam), PASP, 135, 028001

    2023

  45. [45]

    2023, Data from the JWST Advanced Deep Extragalactic Survey (JADES), MAST Release

    Rieke, M., et al. 2023, Data from the JWST Advanced Deep Extragalactic Survey (JADES), MAST Release

    2023

  46. [46]

    R., et al

    Rigby, J. R., et al. 2023, The Science Performance of the James Webb Space Telescope: Early Results, ApJ, 943, 38

    2023

  47. [47]

    Ronneberger, O., et al. T. 2015, U-Net: Convolutional Networks for Biomedical Image Segmentation, in Proc. MICCAI, 234

    2015

  48. [48]

    2025, Detection of Anisotropies in the Circumgalactic Medium of Disk Galaxies: Supermassive Black Hole Activity or Star Formation-driven Outflows?, arXiv:2501.18681

    Sacchi, A. 2025, Detection of Anisotropies in the Circumgalactic Medium of Disk Galaxies: Supermassive Black Hole Activity or Star Formation-driven Outflows?, arXiv:2501.18681

    work page arXiv 2025

  49. [49]

    M., Kolesnikov, I., de Carvalho, R

    Sampaio, V.M. 2026, Morphology across cosmic time: assessing the evolution and interplay of disk and bulge-dominated galaxies in the CANDELS survey, arXiv:2506.12205

    work page arXiv 2026

  50. [50]

    2005, The classification of galaxies: Early history and ongoing developments, Annu

    Sandage, A. 2005, The classification of galaxies: Early history and ongoing developments, Annu. Rev. Astron. Astrophys., 43, 581

    2005

  51. [51]

    2007, The Cosmic Evolution Survey (COSMOS): Overview, ApJS, 172, 1

    Scoville, N., et al. 2007, The Cosmic Evolution Survey (COSMOS): Overview, ApJS, 172, 1

    2007

  52. [52]

    2026, The MAGPI Survey: co-evolution of baryons and dark matter in star-forming disk-like galaxies at 0.1 < z < 0.85,, arXiv:2601.12315

    Sharma, G., et al. 2026, The MAGPI Survey: co-evolution of baryons and dark matter in star-forming disk-like galaxies at 0.1 < z < 0.85,, arXiv:2601.12315

    work page arXiv 2026

  53. [53]

    2025, COSMOS2025: The COSMOS-Web galaxy catalog, A&A, 704, A339

    Shuntov, M., et al. 2025, COSMOS2025: The COSMOS-Web galaxy catalog, A&A, 704, A339

    2025

  54. [54]

    S., & Gallagher, J

    Sparke, L. S., & Gallagher, J. S. 2000, Galaxies in the Universe: An Introduction, Cambridge Univ. Press

    2000

  55. [55]

    2024, The structure and morphology of galaxies during the Epoch of Reionization revealed by JWST, ApJ, 960, 104

    Sun, W., et al. 2024, The structure and morphology of galaxies during the Epoch of Reionization revealed by JWST, ApJ, 960, 104

    2024

  56. [56]

    2025, Automatic machine learning framework to study AGN host morphology, ApJ, 981, 5

    Tian, C., et al. 2025, Automatic machine learning framework to study AGN host morphology, ApJ, 981, 5

    2025

  57. [57]

    2026, Structural Evolution of Stellar Disks from z 3 to the Present with JWST, MNRAS, in press

    Tsukui, T., Yoachim, P., Comerón, S., et al. 2026, Structural Evolution of Stellar Disks from z 3 to the Present with JWST, MNRAS, in press

    2026

  58. [58]

    2014, 3D-HST + CANDELS: Evolution of the size–mass distribution since z=3 , ApJ, 788, 28

    van der Wel, A., et al. 2014, 3D-HST + CANDELS: Evolution of the size–mass distribution since z=3 , ApJ, 788, 28

    2014

  59. [59]

    2024, On the Nature of Disks at High Redshift Seen by JWST/CEERS with Contrastive Learning and Cosmological Simulations, ApJ, 961, 51

    Vega-Ferrero, J., Huertas-Company, M., Costantin, L., et al. 2024, On the Nature of Disks at High Redshift Seen by JWST/CEERS with Contrastive Learning and Cosmological Simulations, ApJ, 961, 51

    2024

  60. [60]

    2022, Galaxy Zoo DECaLS: Detailed visual morphology for 314,000 galaxies, MNRAS, 509, 3966

    Walmsley, M., et al. 2022, Galaxy Zoo DECaLS: Detailed visual morphology for 314,000 galaxies, MNRAS, 509, 3966

    2022

  61. [61]

    2021, General E(2)-Equivariant Steerable CNNs, in Advances in Neural Information Processing Systems (NeurIPS), 14334

    Weiler, M., & Cesa, G. 2021, General E(2)-Equivariant Steerable CNNs, in Advances in Neural Information Processing Systems (NeurIPS), 14334

    2021

  62. [62]

    W., et al

    Willett, K. W., et al. 2013, Galaxy Zoo 2: detailed morphological classifications for 304,122 SDSS galaxies, MNRAS, 435, 2835

    2013

  63. [63]

    2024, A galaxy image augmentation method using few-shot learning and GANs, RAA, 24, 035015

    Yao, Y., et al. 2024, A galaxy image augmentation method using few-shot learning and GANs, RAA, 24, 035015

    2024

  64. [64]

    2026, Clump-like structures in high-redshift galaxies from JADES, arXiv:2601.15965

    Zhu, Y., et al. 2026, Clump-like structures in high-redshift galaxies from JADES, arXiv:2601.15965

    work page arXiv 2026