{"ID":2863178,"CreatedAt":"2026-06-01T04:54:23.091178241Z","UpdatedAt":"2026-06-01T04:54:23.091178241Z","DeletedAt":null,"paper_url":"https://arxiv.org/abs/2509.24134","arxiv_id":"2509.24134","title":"ASTROCO: Self-Supervised Conformer-Style Transformers for Light-Curve Embeddings","abstract":"We present AstroCo, a Conformer-style encoder for irregular stellar light curves. By combining attention with depthwise convolutions and gating, AstroCo captures both global dependencies and local features. On MACHO R-band, AstroCo outperforms Astromer v1 and v2, yielding 70 percent and 61 percent lower error respectively and a relative macro-F1 gain of about 7 percent, while producing embeddings that transfer effectively to few-shot classification. These results highlight AstroCo's potential as a strong and label-efficient foundation for time-domain astronomy.","short_abstract":"We present AstroCo, a Conformer-style encoder for irregular stellar light curves. By combining attention with depthwise convolutions and gating, AstroCo captures both global dependencies and local features. On MACHO R-band, AstroCo outperforms Astromer v1 and v2, yielding 70 percent and 61 percent lower error respectiv...","url_abs":"https://arxiv.org/abs/2509.24134","url_pdf":"https://arxiv.org/pdf/2509.24134v1","authors":"[\"Antony Tan\",\"Pavlos Protopapas\",\"Martina Cádiz-Leyton\",\"Guillermo Cabrera-Vives\",\"Cristobal Donoso-Oliva\",\"Ignacio Becker\"]","published":"2025-09-29T00:09:23Z","proceeding":"astro-ph.IM","tasks":"[\"astro-ph.IM\",\"cs.AI\",\"cs.LG\"]","methods":"[\"Transformer\"]","has_code":false}