Galaxy Morphology Classification via Deep Semisupervised Learning with Limited Labeled Data
Galaxy morphology classification plays a crucial role in understanding the structure and evolution of the Universe. With galaxy observation data growing exponentially, machine learning has become a core technology for this classification task. However, traditional machine learning methods predominan...
Saved in:
| Main Authors: | , , , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
IOP Publishing
2025-01-01
|
| Series: | The Astrophysical Journal Supplement Series |
| Subjects: | |
| Online Access: | https://doi.org/10.3847/1538-4365/addb4c |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1849417399574462464 |
|---|---|
| author | Zhijian Luo Jianzhen Chen Zhu Chen Shaohua Zhang Liping Fu Hubing Xiao Chenggang Shu |
| author_facet | Zhijian Luo Jianzhen Chen Zhu Chen Shaohua Zhang Liping Fu Hubing Xiao Chenggang Shu |
| author_sort | Zhijian Luo |
| collection | DOAJ |
| description | Galaxy morphology classification plays a crucial role in understanding the structure and evolution of the Universe. With galaxy observation data growing exponentially, machine learning has become a core technology for this classification task. However, traditional machine learning methods predominantly rely on supervised learning frameworks, and their dependence on large amounts of labeled samples limits practical applications. To address this challenge, we propose an innovative hybrid semisupervised model, Wasserstein GAN for galaxy classification (GC-SWGAN), designed to tackle galaxy morphology classification under conditions of limited labeled data. This model integrates semisupervised generative adversarial networks (GANs) with Wasserstein GAN with gradient penalty, establishing a multitask learning framework. Within this framework, the discriminator and classifier are designed independently while sharing part of the architecture. By collaborating with the generator, the model significantly enhances both classification performance and sample generation capabilities, while also improving convergence and stability during training. Experimental results demonstrate that, on the Galaxy10 DECaLS data set, GC-SWGAN achieves comparable or even superior classification accuracy (exceeding 75%) using only one-fifth of the labeled samples typically required by conventional fully supervised methods. Under identical labeled conditions, the model displays excellent generalization performance, attaining approximately 84% classification accuracy. Notably, in extreme scenarios where only 10% of the data is labeled, GC-SWGAN still achieves high classification accuracy (over 68%), fully demonstrating its stability and effectiveness in low-labeled data environments. Furthermore, galaxy images generated by GC-SWGAN are visually similar to real samples. Our approach provides a new solution to the challenge of needing to manually label large amounts of data in astronomy. |
| format | Article |
| id | doaj-art-e3945928ca2e4d1dabb1d47ba4280040 |
| institution | Kabale University |
| issn | 0067-0049 |
| language | English |
| publishDate | 2025-01-01 |
| publisher | IOP Publishing |
| record_format | Article |
| series | The Astrophysical Journal Supplement Series |
| spelling | doaj-art-e3945928ca2e4d1dabb1d47ba42800402025-08-20T03:32:51ZengIOP PublishingThe Astrophysical Journal Supplement Series0067-00492025-01-0127911710.3847/1538-4365/addb4cGalaxy Morphology Classification via Deep Semisupervised Learning with Limited Labeled DataZhijian Luo0https://orcid.org/0009-0009-1617-8747Jianzhen Chen1Zhu Chen2https://orcid.org/0000-0002-2326-0476Shaohua Zhang3https://orcid.org/0000-0001-8485-2814Liping Fu4https://orcid.org/0000-0003-0688-8445Hubing Xiao5https://orcid.org/0000-0001-8244-1229Chenggang Shu6Shanghai Key Lab for Astrophysics, Shanghai Normal University , Shanghai 200234, People’s Republic of China ; zjluo@shnu.edu.cnShanghai Key Lab for Astrophysics, Shanghai Normal University , Shanghai 200234, People’s Republic of China ; zjluo@shnu.edu.cnShanghai Key Lab for Astrophysics, Shanghai Normal University , Shanghai 200234, People’s Republic of China ; zjluo@shnu.edu.cnShanghai Key Lab for Astrophysics, Shanghai Normal University , Shanghai 200234, People’s Republic of China ; zjluo@shnu.edu.cnShanghai Key Lab for Astrophysics, Shanghai Normal University , Shanghai 200234, People’s Republic of China ; zjluo@shnu.edu.cn; Center for Astronomy and Space Sciences, China Three Gorges University , Yichang 443000, People’s Republic of ChinaShanghai Key Lab for Astrophysics, Shanghai Normal University , Shanghai 200234, People’s Republic of China ; zjluo@shnu.edu.cnShanghai Key Lab for Astrophysics, Shanghai Normal University , Shanghai 200234, People’s Republic of China ; zjluo@shnu.edu.cnGalaxy morphology classification plays a crucial role in understanding the structure and evolution of the Universe. With galaxy observation data growing exponentially, machine learning has become a core technology for this classification task. However, traditional machine learning methods predominantly rely on supervised learning frameworks, and their dependence on large amounts of labeled samples limits practical applications. To address this challenge, we propose an innovative hybrid semisupervised model, Wasserstein GAN for galaxy classification (GC-SWGAN), designed to tackle galaxy morphology classification under conditions of limited labeled data. This model integrates semisupervised generative adversarial networks (GANs) with Wasserstein GAN with gradient penalty, establishing a multitask learning framework. Within this framework, the discriminator and classifier are designed independently while sharing part of the architecture. By collaborating with the generator, the model significantly enhances both classification performance and sample generation capabilities, while also improving convergence and stability during training. Experimental results demonstrate that, on the Galaxy10 DECaLS data set, GC-SWGAN achieves comparable or even superior classification accuracy (exceeding 75%) using only one-fifth of the labeled samples typically required by conventional fully supervised methods. Under identical labeled conditions, the model displays excellent generalization performance, attaining approximately 84% classification accuracy. Notably, in extreme scenarios where only 10% of the data is labeled, GC-SWGAN still achieves high classification accuracy (over 68%), fully demonstrating its stability and effectiveness in low-labeled data environments. Furthermore, galaxy images generated by GC-SWGAN are visually similar to real samples. Our approach provides a new solution to the challenge of needing to manually label large amounts of data in astronomy.https://doi.org/10.3847/1538-4365/addb4cGalaxiesConvolutional neural networksGround-based astronomyAstronomy data modelingAstronomy data analysisComputational astronomy |
| spellingShingle | Zhijian Luo Jianzhen Chen Zhu Chen Shaohua Zhang Liping Fu Hubing Xiao Chenggang Shu Galaxy Morphology Classification via Deep Semisupervised Learning with Limited Labeled Data The Astrophysical Journal Supplement Series Galaxies Convolutional neural networks Ground-based astronomy Astronomy data modeling Astronomy data analysis Computational astronomy |
| title | Galaxy Morphology Classification via Deep Semisupervised Learning with Limited Labeled Data |
| title_full | Galaxy Morphology Classification via Deep Semisupervised Learning with Limited Labeled Data |
| title_fullStr | Galaxy Morphology Classification via Deep Semisupervised Learning with Limited Labeled Data |
| title_full_unstemmed | Galaxy Morphology Classification via Deep Semisupervised Learning with Limited Labeled Data |
| title_short | Galaxy Morphology Classification via Deep Semisupervised Learning with Limited Labeled Data |
| title_sort | galaxy morphology classification via deep semisupervised learning with limited labeled data |
| topic | Galaxies Convolutional neural networks Ground-based astronomy Astronomy data modeling Astronomy data analysis Computational astronomy |
| url | https://doi.org/10.3847/1538-4365/addb4c |
| work_keys_str_mv | AT zhijianluo galaxymorphologyclassificationviadeepsemisupervisedlearningwithlimitedlabeleddata AT jianzhenchen galaxymorphologyclassificationviadeepsemisupervisedlearningwithlimitedlabeleddata AT zhuchen galaxymorphologyclassificationviadeepsemisupervisedlearningwithlimitedlabeleddata AT shaohuazhang galaxymorphologyclassificationviadeepsemisupervisedlearningwithlimitedlabeleddata AT lipingfu galaxymorphologyclassificationviadeepsemisupervisedlearningwithlimitedlabeleddata AT hubingxiao galaxymorphologyclassificationviadeepsemisupervisedlearningwithlimitedlabeleddata AT chenggangshu galaxymorphologyclassificationviadeepsemisupervisedlearningwithlimitedlabeleddata |