{
  "abstract": "Auroral spectrographs such as the Auroral Spectrograph In Skibotn (ASIS) record hundreds of thousands of emission spectra, but only a few hundred can be labelled by an expert. To exploit the rest, we pretrain a 1D Vision Transformer with a masked autoencoder on 223,000 unlabelled spectra. Without labels, its representation recovers the emission-line intensity ratios that physicists use to diagnose the precipitating particles (R^2 0.91 vs. 0.77 for an untrained control) and, under one linear probe, classifies as well as 13 features designed by experts. Fine-tuned, the model outperforms the previous supervised auroral classifier on its own benchmark (macro-AP 88.5 vs. 77.8), reaches 0.870 mAP, and exceeds the same architecture trained from scratch by +0.159 with 10% of the labels; attribution shows that it uses both N2+ bands. Could an existing pretrained model replace it? Two astronomical spectral foundation models and a time-series model transfer according to their spectral window: SpectraFM, trained in the infrared, falls below the untrained control, whereas SpecFormer, trained in the optical, approaches in-domain pretraining without reaching it.",
  "abstract_shown": true,
  "affiliations": [],
  "article_type": null,
  "arxiv": "2609.31206",
  "author_total": 3,
  "authors": [
    {
      "name": "Matthieu Le Lain"
    },
    {
      "name": "Gaël Cessateur"
    },
    {
      "name": "Sébastien Lefèvre"
    }
  ],
  "bibtex": "https://research.draconitic.com/paper/2s0wff1brp2cp.bib",
  "cited_by": 0,
  "doi": "10.48550/arxiv.2609.31206",
  "fulltext_url": null,
  "id": "2s0wff1brp2cp",
  "issns": [],
  "landing_url": "https://arxiv.org/abs/2609.31206",
  "language": null,
  "licence": "LicenseRef-arXiv-nonexclusive-1.0",
  "licence_label": "arXiv licence",
  "licence_source": "article",
  "licence_url": "http://arxiv.org/licenses/nonexclusive-distrib/1.0/",
  "metadata_licence": "CC0-1.0",
  "native_subjects": [
    [
      "arxiv",
      "cs.LG"
    ],
    [
      "arxiv",
      "physics.space-ph"
    ],
    [
      "openalex",
      "3103|Ionosphere and magnetosphere dynamics"
    ]
  ],
  "openalex": "W7214575328",
  "pdf_url": "https://arxiv.org/pdf/2609.31206",
  "pmcid": null,
  "pmid": null,
  "published": "2026-09-25",
  "publisher": null,
  "ris": "https://research.draconitic.com/paper/2s0wff1brp2cp.ris",
  "sources": [
    {
      "first_seen": "2026-10-30T12:24:52Z",
      "native_id": "2609.31206",
      "source": "arxiv",
      "url": "https://arxiv.org/abs/2609.31206"
    }
  ],
  "subjects": [
    {
      "field": "astronomy",
      "origin": "arxiv:physics.space-ph",
      "subfield": "stellar"
    }
  ],
  "title": "Self-Supervised Representation Learning: From Spectral Foundation Models to Auroral Emission Spectra",
  "title_text": "Self-Supervised Representation Learning: From Spectral Foundation Models to Auroral Emission Spectra",
  "type": "preprint",
  "updated_at": "2026-10-30T12:24:52Z",
  "url": "https://research.draconitic.com/paper/2s0wff1brp2cp",
  "venue": "arXiv",
  "versions": []
}