ASTROCO: Self-Supervised Conformer-Style Transformers for Light-Curve Embeddings
Fuente:
arXiv
Salvato in:
| Autori principali: | , , , , , |
|---|---|
| Natura: | Preprint |
| Pubblicazione: |
2025
|
| Soggetti: | |
| Accesso online: | |
| Tags: |
Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
|
| _version_ | 1866908563842007040 |
|---|---|
| author | Tan, Antony Protopapas, Pavlos Cádiz-Leyton, Martina Cabrera-Vives, Guillermo Donoso-Oliva, Cristobal Becker, Ignacio |
| author_facet | Tan, Antony Protopapas, Pavlos Cádiz-Leyton, Martina Cabrera-Vives, Guillermo Donoso-Oliva, Cristobal Becker, Ignacio |
| contents | 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. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2509_24134 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | ASTROCO: Self-Supervised Conformer-Style Transformers for Light-Curve Embeddings Tan, Antony Protopapas, Pavlos Cádiz-Leyton, Martina Cabrera-Vives, Guillermo Donoso-Oliva, Cristobal Becker, Ignacio Instrumentation and Methods for Astrophysics Artificial Intelligence Machine Learning 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. |
| title | ASTROCO: Self-Supervised Conformer-Style Transformers for Light-Curve Embeddings |
| topic | Instrumentation and Methods for Astrophysics Artificial Intelligence Machine Learning |
| url | https://arxiv.org/abs/2509.24134 |