Transfer learning for radio galaxy classificationCitation formats

Standard

Transfer learning for radio galaxy classification. / Tang, Hongming; Scaife, Anna M. M.; Leahy, J. P.

In: Monthly Notices of the Royal Astronomical Society, Vol. 488, 25.07.2019, p. 3358-3375.

Research output: Contribution to journalArticlepeer-review

Harvard

Tang, H, Scaife, AMM & Leahy, JP 2019, 'Transfer learning for radio galaxy classification', Monthly Notices of the Royal Astronomical Society, vol. 488, pp. 3358-3375. https://doi.org/10.1093/mnras/stz1883

APA

Tang, H., Scaife, A. M. M., & Leahy, J. P. (2019). Transfer learning for radio galaxy classification. Monthly Notices of the Royal Astronomical Society, 488, 3358-3375. https://doi.org/10.1093/mnras/stz1883

Vancouver

Tang H, Scaife AMM, Leahy JP. Transfer learning for radio galaxy classification. Monthly Notices of the Royal Astronomical Society. 2019 Jul 25;488:3358-3375. https://doi.org/10.1093/mnras/stz1883

Author

Tang, Hongming ; Scaife, Anna M. M. ; Leahy, J. P. / Transfer learning for radio galaxy classification. In: Monthly Notices of the Royal Astronomical Society. 2019 ; Vol. 488. pp. 3358-3375.

Bibtex

@article{d052cc63d90741deae38e94a7b55ff23,
title = "Transfer learning for radio galaxy classification",
abstract = " In the context of radio galaxy classification, most state-of-the-art neural network algorithms have been focused on single survey data. The question of whether these trained algorithms have cross-survey identification ability or can be adapted to develop classification networks for future surveys is still unclear. One possible solution to address this issue is transfer learning, which re-uses elements of existing machine learning models for different applications. Here we present radio galaxy classification based on a 13-layer Deep Convolutional Neural Network (DCNN) using transfer learning methods between different radio surveys. We find that our machine learning models trained from a random initialization achieve accuracies comparable to those found elsewhere in the literature. When using transfer learning methods, we find that inheriting model weights pre-trained on FIRST images can boost model performance when re-training on lower resolution NVSS data, but that inheriting pre-trained model weights from NVSS and re-training on FIRST data impairs the performance of the classifier. We consider the implication of these results in the context of future radio surveys planned for next-generation radio telescopes such as ASKAP, MeerKAT, and SKA1-MID. ",
keywords = "astro-ph.IM",
author = "Hongming Tang and Scaife, {Anna M. M.} and Leahy, {J. P.}",
year = "2019",
month = jul,
day = "25",
doi = "10.1093/mnras/stz1883",
language = "English",
volume = "488",
pages = "3358--3375",
journal = "Royal Astronomical Society. Monthly Notices",
issn = "1365-2966",
publisher = "Oxford University Press",

}

RIS

TY - JOUR

T1 - Transfer learning for radio galaxy classification

AU - Tang, Hongming

AU - Scaife, Anna M. M.

AU - Leahy, J. P.

PY - 2019/7/25

Y1 - 2019/7/25

N2 - In the context of radio galaxy classification, most state-of-the-art neural network algorithms have been focused on single survey data. The question of whether these trained algorithms have cross-survey identification ability or can be adapted to develop classification networks for future surveys is still unclear. One possible solution to address this issue is transfer learning, which re-uses elements of existing machine learning models for different applications. Here we present radio galaxy classification based on a 13-layer Deep Convolutional Neural Network (DCNN) using transfer learning methods between different radio surveys. We find that our machine learning models trained from a random initialization achieve accuracies comparable to those found elsewhere in the literature. When using transfer learning methods, we find that inheriting model weights pre-trained on FIRST images can boost model performance when re-training on lower resolution NVSS data, but that inheriting pre-trained model weights from NVSS and re-training on FIRST data impairs the performance of the classifier. We consider the implication of these results in the context of future radio surveys planned for next-generation radio telescopes such as ASKAP, MeerKAT, and SKA1-MID.

AB - In the context of radio galaxy classification, most state-of-the-art neural network algorithms have been focused on single survey data. The question of whether these trained algorithms have cross-survey identification ability or can be adapted to develop classification networks for future surveys is still unclear. One possible solution to address this issue is transfer learning, which re-uses elements of existing machine learning models for different applications. Here we present radio galaxy classification based on a 13-layer Deep Convolutional Neural Network (DCNN) using transfer learning methods between different radio surveys. We find that our machine learning models trained from a random initialization achieve accuracies comparable to those found elsewhere in the literature. When using transfer learning methods, we find that inheriting model weights pre-trained on FIRST images can boost model performance when re-training on lower resolution NVSS data, but that inheriting pre-trained model weights from NVSS and re-training on FIRST data impairs the performance of the classifier. We consider the implication of these results in the context of future radio surveys planned for next-generation radio telescopes such as ASKAP, MeerKAT, and SKA1-MID.

KW - astro-ph.IM

U2 - 10.1093/mnras/stz1883

DO - 10.1093/mnras/stz1883

M3 - Article

VL - 488

SP - 3358

EP - 3375

JO - Royal Astronomical Society. Monthly Notices

JF - Royal Astronomical Society. Monthly Notices

SN - 1365-2966

ER -