ATAT: Astronomical Transformer for time series and Tabular data

Simple item page
dc.article.numberA289
dc.catalogadorjwg
dc.contributor.authorCabrera-Vives G.
dc.contributor.authorMoreno-Cartagena D.
dc.contributor.authorAstorga N.
dc.contributor.authorReyes-Jainaga I.
dc.contributor.authorForster F.
dc.contributor.authorHuijse P.
dc.contributor.authorArredondo J.
dc.contributor.authorMunoz Arancibia A.M.
dc.contributor.authorBayo A.
dc.contributor.authorCatelan, Marcio
dc.contributor.authorEstevez P.A.
dc.contributor.authorSanchez-Saez P.
dc.contributor.authorAlvarez A.
dc.contributor.authorCastellanos P.
dc.contributor.authorGallardo P.
dc.contributor.authorMoya A.
dc.contributor.authorRodriguez-Mancini D.
dc.date.accessioned2025-03-17T13:48:43Z
dc.date.available2025-03-17T13:48:43Z
dc.date.issued2024
dc.description.abstractThe advent of next-generation survey instruments, such as the Vera C. Rubin Observatory and its Legacy Survey of Space and Time (LSST), is opening a window for new research in time-domain astronomy. The Extended LSST Astronomical Time-Series Classification Challenge (ELAsTiCC) was created to test the capacity of brokers to deal with a simulated LSST stream. Aims. Our aim is to develop a next-generation model for the classification of variable astronomical objects. We describe ATAT, the Astronomical Transformer for time series And Tabular data, a classification model conceived by the ALeRCE alert broker to classify light curves from next-generation alert streams. ATAT was tested in production during the first round of the ELAsTiCC campaigns. Methods. ATAT consists of two transformer models that encode light curves and features using novel time modulation and quantile feature tokenizer mechanisms, respectively. ATAT was trained on different combinations of light curves, metadata, and features calculated over the light curves. We compare ATAT against the current ALeRCE classifier, a balanced hierarchical random forest (BHRF) trained on human-engineered features derived from light curves and metadata. Results. When trained on light curves and metadata, ATAT achieves a macro F1 score of 82.9 ± 0.4 in 20 classes, outperforming the BHRF model trained on 429 features, which achieves a macro F1 score of 79.4 ± 0.1. Conclusions. The use of transformer multimodal architectures, combining light curves and tabular data, opens new possibilities for classifying alerts from a new generation of large etendue telescopes, such as the Vera C. Rubin Observatory, in real-world brokering scenarios.
dc.description.funderFONDEQUIP
dc.description.funderU.S. Department of Energy-supported Dark Energy Science Collaboration
dc.description.funderNational Agency for Research and Development
dc.description.funderLawrence Berkeley National Laboratory
dc.description.funderU.S. Department of Energy Office of Science
dc.description.funderFONDECYT
dc.description.funderANID
dc.description.funderUniversidad Austral de Chile
dc.fuente.origenORCID
dc.identifier.doi10.1051/0004-6361/202449475
dc.identifier.issn14320746 00046361
dc.identifier.scopusidSCOPUS_ID:85205011730
dc.identifier.urihttps://repositorio.uc.cl/handle/11534/102655
dc.information.autorucInstituto de Astrofísica; Catelan, Marcio; 0000-0001-6003-8877; 1001556
dc.language.isoen
dc.nota.accesocontenido completo
dc.publisherEDP Sciences
dc.revistaAstronomy and Astrophysics
dc.rightsacceso abierto
dc.subjectmethods: data analysis
dc.subjectmethods: statistical
dc.subjectstars: variables: general
dc.subjectsupernovae: general
dc.subjectsurveys
dc.subject.ddc520
dc.subject.deweyAstronomíaes_ES
dc.titleATAT: Astronomical Transformer for time series and Tabular data
dc.typeartículo
dc.volumen689
sipa.codpersvinculados1001556
sipa.trazabilidadORCID;2025-03-03
Files
Original bundle
Now showing 1 - 1 of 1
Thumbnail Image
Name:
ATAT Astronomical Transformer for time series and Tabular data.pdf
Size:
695.88 KB
Format:
Adobe Portable Document Format
Description:
Download
Collections
Artículos de revistas