Feature-Based and Deep Learning Methods for Astronomical Light Curve Classification
Abstract
We compare feature-based machine learning and end-to-end deep learning for variable star classification using the Zwicky Transient Facility (ZTF) StarEmbed dataset, which contains about 40,000 objects across seven imbalanced classes. Classical models trained on physically motivated features consistently outperform deep networks trained directly on raw light curves. We find that limited labeled data contributes to the performance gap, while the multivariate, irregularly sampled, and morphologically diverse nature of light curves poses additional challenges for deep learning models. While deep learning holds promise for learning representations directly from observations, our results indicate that current models fail to generalize with the available training set size. These findings motivate further investigation of the benefits of scaling labeled datasets and self-supervised pretraining on large unlabeled surveys. This work establishes a quantitative baseline for future research and highlights the need for further innovations to unlock the potential of deep learning in astronomical time series analysis.
Related
- Contrastive Astromer: Two-Stage Contrastive Representation Learning for Astronomical Light Curve Classification
- Multi-Method Machine Learning Classification of Astronomical Sources with Rubin Data Previews: a Spectroscopy-Free Approach with DP1 and DP2
- The NEMESIS general young stellar object catalogue. I. Supervised classification with deep learning methods
- HyQ-LC: Improving Light-Curve Classification with Hybrid Quantum--Classical Learning
- The NEMESIS general YSO catalogue - I. Supervised classification with deep learning methods
BibTeX
@misc{baymakanov2026featurebased,
title = {{Feature-Based and Deep Learning Methods for Astronomical Light Curve Classification}},
author = {V. Baymakanov and Kh. Khamitov and A. Raimova and A. Isaeva and R. Taran and M. Zedan and M. Kornilov and M. Demianenko and K. Malanchev and M. Hushchyn},
howpublished = {arXiv},
year = {2026},
month = nov,
doi = {10.48550/arxiv.2611.04664},
eprint = {2611.04664},
archivePrefix = {arXiv},
url = {https://arxiv.org/abs/2611.04664},
}
RIS
TY - UNPB
TI - Feature-Based and Deep Learning Methods for Astronomical Light Curve Classification
AU - V. Baymakanov
AU - Kh. Khamitov
AU - A. Raimova
AU - A. Isaeva
AU - R. Taran
AU - M. Zedan
AU - M. Kornilov
AU - M. Demianenko
AU - K. Malanchev
AU - M. Hushchyn
PY - 2026
DA - 2026/11/03
PB - arXiv
DO - 10.48550/arxiv.2611.04664
UR - https://arxiv.org/abs/2611.04664
AB - We compare feature-based machine learning and end-to-end deep learning for variable star classification using the Zwicky Transient Facility (ZTF) StarEmbed dataset, which contains about 40,000 objects across seven imbalanced classes. Classical models trained on physically motivated features consistently outperform deep networks trained directly on raw light curves. We find that limited labeled data contributes to the performance gap, while the multivariate, irregularly sampled, and morphologically diverse nature of light curves poses additional challenges for deep learning models. While deep learning holds promise for learning representations directly from observations, our results indicate that current models fail to generalize with the available training set size. These findings motivate further investigation of the benefits of scaling labeled datasets and self-supervised pretraining on large unlabeled surveys. This work establishes a quantitative baseline for future research and highlights the need for further innovations to unlock the potential of deep learning in astronomical time series analysis.
ER -