Efficient galaxy classification through pretraining
Deep learning has increasingly been applied to supervised learning tasks in astronomy, such as classifying images of galaxies based on their apparent shape (i.e., galaxy morphology classification) to gain insight regarding the evolution of galaxies. In this work, we examine the effect of pretraining...
| Main Authors: | , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Frontiers Media S.A.
2023-08-01
|
| Series: | Frontiers in Astronomy and Space Sciences |
| Subjects: | |
| Online Access: | https://www.frontiersin.org/articles/10.3389/fspas.2023.1197358/full |
| _version_ | 1797745476213145600 |
|---|---|
| author | Jesse Schneider David C. Stenning Lloyd T. Elliott |
| author_facet | Jesse Schneider David C. Stenning Lloyd T. Elliott |
| author_sort | Jesse Schneider |
| collection | DOAJ |
| description | Deep learning has increasingly been applied to supervised learning tasks in astronomy, such as classifying images of galaxies based on their apparent shape (i.e., galaxy morphology classification) to gain insight regarding the evolution of galaxies. In this work, we examine the effect of pretraining on the performance of the classical AlexNet convolutional neural network (CNN) in classifying images of 14,034 galaxies from the Sloan Digital Sky Survey Data Release 4. Pretraining involves designing and training CNNs on large labeled image datasets unrelated to astronomy, which takes advantage of the vast amounts of such data available compared to the relatively small amount of labeled galaxy images. We show a statistically significant benefit of using pretraining, both in terms of improved overall classification success and reduced computational cost to achieve such performance. |
| first_indexed | 2024-03-12T15:23:47Z |
| format | Article |
| id | doaj.art-750514d99533456488b162c0b29500d4 |
| institution | Directory Open Access Journal |
| issn | 2296-987X |
| language | English |
| last_indexed | 2024-03-12T15:23:47Z |
| publishDate | 2023-08-01 |
| publisher | Frontiers Media S.A. |
| record_format | Article |
| series | Frontiers in Astronomy and Space Sciences |
| spelling | doaj.art-750514d99533456488b162c0b29500d42023-08-10T16:14:15ZengFrontiers Media S.A.Frontiers in Astronomy and Space Sciences2296-987X2023-08-011010.3389/fspas.2023.11973581197358Efficient galaxy classification through pretrainingJesse SchneiderDavid C. StenningLloyd T. ElliottDeep learning has increasingly been applied to supervised learning tasks in astronomy, such as classifying images of galaxies based on their apparent shape (i.e., galaxy morphology classification) to gain insight regarding the evolution of galaxies. In this work, we examine the effect of pretraining on the performance of the classical AlexNet convolutional neural network (CNN) in classifying images of 14,034 galaxies from the Sloan Digital Sky Survey Data Release 4. Pretraining involves designing and training CNNs on large labeled image datasets unrelated to astronomy, which takes advantage of the vast amounts of such data available compared to the relatively small amount of labeled galaxy images. We show a statistically significant benefit of using pretraining, both in terms of improved overall classification success and reduced computational cost to achieve such performance.https://www.frontiersin.org/articles/10.3389/fspas.2023.1197358/fullconvolutional neural networksmachine learninggalaxy morphologyastrostatisticstransfer learningpretraining |
| spellingShingle | Jesse Schneider David C. Stenning Lloyd T. Elliott Efficient galaxy classification through pretraining Frontiers in Astronomy and Space Sciences convolutional neural networks machine learning galaxy morphology astrostatistics transfer learning pretraining |
| title | Efficient galaxy classification through pretraining |
| title_full | Efficient galaxy classification through pretraining |
| title_fullStr | Efficient galaxy classification through pretraining |
| title_full_unstemmed | Efficient galaxy classification through pretraining |
| title_short | Efficient galaxy classification through pretraining |
| title_sort | efficient galaxy classification through pretraining |
| topic | convolutional neural networks machine learning galaxy morphology astrostatistics transfer learning pretraining |
| url | https://www.frontiersin.org/articles/10.3389/fspas.2023.1197358/full |
| work_keys_str_mv | AT jesseschneider efficientgalaxyclassificationthroughpretraining AT davidcstenning efficientgalaxyclassificationthroughpretraining AT lloydtelliott efficientgalaxyclassificationthroughpretraining |