Comparative Study of Activation Functions and Image Resolution on ResNet-34 for Spiral Galaxy Spin Classification
DOI:
https://doi.org/10.21009/SPEKTRA.103.03Keywords:
spiral galaxies, spin direction, ResNet, ReLU, image resolution, deep learningAbstract
This study investigates the application of the Residual Network (ResNet-34) architecture for classifying spiral galaxy spin directions, specifically focusing on the comparative performance of activation functions and cross-dataset generalizability using data derived from the Dark Energy Spectroscopic Instrument Legacy Survey (DESI LS) and the Hyper Suprime Cam Subaru Strategic Program (HSC-SSP) surveys. The methodology ensures robustness by training each model configuration across 10 independent runs. The results demonstrate the clear superiority of the Rectified Linear Unit (ReLU) over the Hyperbolic Tangent (Tanh); ReLU-based models achieved a mean peak accuracy of 94.7% and required only less than 60 epochs to converge, significantly faster than Tanh's 120 epochs. Crucially, we found that models trained on lower-resolution DESI LS images exhibited superior robustness and generalizability compared to high-resolution-trained models, suggesting that low-resolution training acts as effective implicit regularization. This research provides critical design recommendations for efficient machine learning pipelines, particularly for upcoming facilities like the 3.8-meter telescope at Timau National Observatory (TNO), ensuring model stability and transferability across diverse survey conditions.
References
C. C. Lin and F. H. Shu, “On the spiral structure of disk galaxies,” Astrophys. J., vol. 140, p. 646, Aug. 1964, doi: 10.1086/147955.
J. Baba, T. R. Saitoh, and K. Wada, “Dynamics of non-steady spiral arms in disk galaxies,” Astrophys. J., vol. 763, no. 1, p. 46, 2013, doi: 10.1088/0004-637X/763/1/46.
C. Dobbs and J. Baba, “Dawes Review 4: Spiral structures in disc galaxies,” Publ. Astron. Soc. Aust., vol. 31, p. e035, 2014, doi: 10.1017/pasa.2014.31.
A. G. Doroshkevich, “Spatial structure of perturbations and origin of galactic rotation in fluctuation theory,” Astrophysics, vol. 6, 1970, doi: 10.1007/BF01001625.
P. J. E. Peebles, “Origin of the angular momentum of galaxies,” Astrophys. J., vol. 155, p. 393, Feb. 1969, doi: 10.1086/149876.
S. D. M. White, “Angular momentum growth in protogalaxies,” Astrophys. J., vol. 286, pp. 38–41, Nov. 1984, doi: 10.1086/162573.
H. Sugai and M. Iye, “A statistical search for correlations of rotational spin angular momentum between galaxies,” Mon. Not. R. Astron. Soc., vol. 276, no. 1, pp. 327–335, Sep. 1995, doi: 10.1093/mnras/276.1.327.
J. Lee and U.-L. Pen, “Cosmic shear from galaxy spins,” Astrophys. J., vol. 532, no. 1, p. L5, 2000, doi: 10.1086/312556.
J. Lee and U.-L. Pen, “Galaxy spin statistics and spin-density correlation,” Astrophys. J., vol. 555, no. 1, p. 106, 2001, doi: 10.1086/321472.
P. Motloch, H.-R. Yu, U.-L. Pen, and Y. Xie, “An observed correlation between galaxy spins and initial conditions,” Nat. Astron., vol. 5, no. 3, pp. 283–288, Jul. 2021, doi: 10.1038/s41550-020-01262-3.
H.-R. Yu et al., “Probing primordial chirality with galaxy spins,” Phys. Rev. Lett., vol. 124, no. 10, p. 101302, Mar. 2020, doi: 10.1103/PhysRevLett.124.101302.
H.-R. Yu, U.-L. Pen, and X. Wang, “Parity-odd neutrino torque detection,” Phys. Rev. D, vol. 99, no. 12, p. 123532, Jun. 2019, doi: 10.1103/PhysRevD.99.123532.
M. Iye, K. Tadaki, and H. Fukumoto, “Spin parity of spiral galaxies. I. Corroborative evidence for trailing spirals,” Astrophys. J., vol. 886, no. 2, p. 133, Nov. 2019, doi: 10.3847/1538-4357/ab4a18.
C. J. Lintott et al., “Galaxy Zoo: Morphologies derived from visual inspection of galaxies from the Sloan Digital Sky Survey,” Mon. Not. R. Astron. Soc., vol. 389, no. 3, pp. 1179–1189, Jul. 2008, doi: 10.1111/j.1365-2966.2008.13689.x.
C. Lintott et al., “Galaxy Zoo 1: Data release of morphological classifications for nearly 900 000 galaxies,” Mon. Not. R. Astron. Soc., vol. 410, no. 1, pp. 166–178, Jan. 2011, doi: 10.1111/j.1365-2966.2010.17432.x.
K. N. Abazajian et al., “The seventh data release of the Sloan Digital Sky Survey,” Astrophys. J. Suppl. Ser., vol. 182, no. 2, p. 543, 2009, doi: 10.1088/0067-0049/182/2/543.
R. Ahumada et al., “The sixteenth data release of the Sloan Digital Sky Surveys,” Astrophys. J. Suppl. Ser., vol. 249, no. 1, p. 3, 2020, doi: 10.3847/1538-4365/ab929e.
A. Dey et al., “Overview of the DESI legacy imaging surveys,” Astron. J., vol. 157, no. 5, p. 168, 2019, doi: 10.3847/1538-3881/ab089d.
H. Aihara et al., “The Hyper Suprime-Cam SSP survey: Overview and survey design,” Publ. Astron. Soc. Jpn., vol. 70, no. SP1, p. S4, Jan. 2018, doi: 10.1093/pasj/psx066.
M. A. Hayat et al., “Self-supervised representation learning for astronomical images,” Astrophys. J. Lett., vol. 911, no. 2, p. L33, 2021, doi: 10.3847/2041-8213/abf2c7.
S. He et al., “Learning to predict the cosmological structure formation,” Proc. Natl. Acad. Sci. U.S.A., vol. 116, no. 28, pp. 13825–13832, 2019, doi: 10.1073/pnas.1821458116.
Ž. Ivezić et al., “LSST: From science drivers to reference design and anticipated data products,” Astrophys. J., vol. 873, no. 2, p. 111, 2019, doi: 10.3847/1538-4357/ab042c.
Euclid Collaboration et al., “Euclid,” Astron. Astrophys., vol. 697, 2025, doi: 10.1051/0004-6361/202450810.
E. S. Mumpuni et al., “Future astronomy facilities in Indonesia,” Nat. Astron., vol. 2, no. 12, pp. 930–932, 2018, doi: 10.1038/s41550-018-0642-6.
T. Szandała, “Review and comparison of commonly used activation functions for deep neural networks,” in Bio-inspired Neurocomputing, A. K. Bhoi et al., Eds. Singapore: Springer, 2021, pp. 203–224, doi: 10.1007/978-981-15-5495-7_11.
K. Tadaki et al., “Spin parity of spiral galaxies II,” Mon. Not. R. Astron. Soc., vol. 496, no. 4, pp. 4276–4286, Aug. 2020, doi: 10.1093/mnras/staa1880.
H. Jia, H.-M. Zhu, and U.-L. Pen, “Galaxy spin classification. I,” Astrophys. J., vol. 943, no. 1, p. 32, 2023, doi: 10.3847/1538-4357/aca8aa.
D. Schlegel et al., “DESI legacy imaging surveys data release 9,” in Proc. AAS Meeting, vol. 237, Jan. 2021, p. 235.03.
H. Aihara et al., “Third data release of the Hyper Suprime-Cam Subaru Strategic Program,” Publ. Astron. Soc. Jpn., vol. 74, no. 2, pp. 247–272, Apr. 2022, doi: 10.1093/pasj/psab122.
K. He and J. Sun, “Convolutional neural networks at constrained time cost,” in Proc. IEEE Conf. Comput. Vis. Pattern Recognit. (CVPR), 2015, pp. 5353–5360, doi: 10.1109/CVPR.2015.7299173.
K. He, X. Zhang, S. Ren, and J. Sun, “Deep residual learning for image recognition,” in Proc. IEEE Conf. Comput. Vis. Pattern Recognit. (CVPR), 2016, pp. 770–778, doi: 10.1109/CVPR.2016.90.
A. Paszke et al., “PyTorch: An imperative style, high-performance deep learning library,” in Adv. Neural Inf. Process. Syst., 2019.
Downloads
Published
How to Cite
Issue
Section
License
Copyright (c) 2025 Hafiz Indra Arwinata, Sultan Hadi Kusuma, Anton Timur Jaelani
This work is licensed under a Creative Commons Attribution 4.0 International License.
SPEKTRA: Jurnal Fisika dan Aplikasinya allow the author(s) to hold the copyright without restrictions and allow the author(s) to retain publishing rights without restrictions. SPEKTRA: Jurnal Fisika dan Aplikasinya CC-BY or an equivalent license as the optimal license for the publication, distribution, use, and reuse of scholarly work. In developing strategy and setting priorities, SPEKTRA: Jurnal Fisika dan Aplikasinya recognize that free access is better than priced access, libre access is better than free access, and libre under CC-BY or the equivalent is better than libre under more restrictive open licenses. We should achieve what we can when we can. We should not delay achieving free in order to achieve libre, and we should not stop with free when we can achieve libre.
SPEKTRA: Jurnal Fisika dan Aplikasinya is licensed under a Creative Commons Attribution 4.0 International License.
You are free to:
Share - copy and redistribute the material in any medium or format
Adapt - remix, transform, and build upon the material for any purpose, even commercially.
The licensor cannot revoke these freedoms as long as you follow the license terms.
E-ISSN 2541-3392 
Focus & Scope 
Editorial Team 
Reviewer Team 
Author Guidelines 
Article Template 
Author Fee 
Publication Ethics 
Plagiarism Policy 
Open Access Policy 
Peer Review Process 
Retraction & Correction 
Licensing & Copyright 
Archiving & Repository 
Contact 
Mendeley 
