Skip to content
DraconiticResearch
Search Sources API
181,686 studies31 sourcesUpdated 4 min ago
PreprintCC BY 4.0arXiv

Self-Supervised Representation Learning for JWST Galaxy Images and Low Resolution Spectra

  • Yang Cheng
  • Morgan Fouesneau
  • Ivelina G. Momcheva
  • Tobias Buck
  • Shengxiu Sun
Open at sourcePDFDOICite

Abstract

A self-supervised multimodal model recovers galaxy stellar masses from sparse JWST data that lack the multi-band coverage required by traditional SED fitting. We train on single-band F150W imaging paired with observed-frame 1-2 μm PRISM spectra and obtain a joint image-spectrum representation that recovers stellar mass at 1<z≤3 with R2 ∼ 0.9 and σNMAD = 0.24 dex; cross-modal CLIP alignment further raises single-modality performance by ΔR2 = +0.04 (imaging) and +0.09 (spectroscopy), narrowing the gap to joint-modality inference. The code and documentation are available at \href{https://github.com/2003cy/neurips2026_ssl_jwst}{GitHub}.

Related

  1. Self-Supervised Representation Learning: From Spectral Foundation Models to Auroral Emission Spectra
    Matthieu Le Lain, Gaël Cessateur, Sébastien Lefèvre
    arXivPreprintarXiv licenceFull textSolar & stellar physicsarXiv
  2. A Low-Dimensional Chemical Representation of High-Resolution Stellar Disk Spectra: Residual Abundance Signatures
    Sarah E. Aquilina, Melissa K. Ness, Sven Buder
    arXivPreprintCC BY 4.0Full textGalaxiesSolar & stellar physicsarXiv
  3. A Low-Dimensional Chemical Representation of High-Resolution Stellar Disk Spectra: Residual Abundance Signatures
    Sarah E Aquilina, Melissa K Ness, Sven Buder
    Monthly Notices of the Royal Astronomical SocietyPeer-reviewedCC BY 4.0Full textSolar & stellar physicsCrossref
  4. Lyman Break Galaxy selection and redshift measurement with supervised contrastive learning
    J Choppin de Janvry, C Yèche, Arjun Dey +55
    HALPreprintLicence not statedGeneral astronomyGeneral physicsHAL
  5. Learning Galaxy Kinematics from Spatially Unresolved Spectra
    Preston G. Waldrop, Dimitrios Psaltis, Feryal Özel +1
    arXivPreprintCC BY 4.0Full textGalaxiesarXiv

Cite

BibTeXRIS
BibTeX
@misc{cheng2026selfsupervised,
  title = {{Self-Supervised Representation Learning for JWST Galaxy Images and Low Resolution Spectra}},
  author = {Yang Cheng and Morgan Fouesneau and Ivelina G. Momcheva and Tobias Buck and Shengxiu Sun},
  howpublished = {arXiv},
  year = {2026},
  month = oct,
  doi = {10.48550/arxiv.2610.31078},
  eprint = {2610.31078},
  archivePrefix = {arXiv},
  url = {https://arxiv.org/abs/2610.31078},
  note = {CC BY 4.0},
}
RIS
TY  - UNPB
TI  - Self-Supervised Representation Learning for JWST Galaxy Images and Low Resolution Spectra
AU  - Yang Cheng
AU  - Morgan Fouesneau
AU  - Ivelina G. Momcheva
AU  - Tobias Buck
AU  - Shengxiu Sun
PY  - 2026
DA  - 2026/10/22
PB  - arXiv
DO  - 10.48550/arxiv.2610.31078
UR  - https://arxiv.org/abs/2610.31078
AB  - A self-supervised multimodal model recovers galaxy stellar masses from sparse JWST data that lack the multi-band coverage required by traditional SED fitting. We train on single-band F150W imaging paired with observed-frame 1-2 μm PRISM spectra and obtain a joint image-spectrum representation that recovers stellar mass at 1<z≤3 with R^2 ∼ 0.9 and σ_NMAD = 0.24 dex; cross-modal CLIP alignment further raises single-modality performance by ΔR^2 = +0.04 (imaging) and +0.09 (spectroscopy), narrowing the gap to joint-modality inference. The code and documentation are available at \href{https://github.com/2003cy/neurips2026_ssl_jwst}{GitHub}.
ER  -