Classifying Complex Faraday Spectra with Convolutional Neural Networks

Brown, Shea; Bergerud, Brandon M.; Costa, Allison H.; Gaensler, B. M.; Isbell, J. W.; LaRocca, Daniel M.; Norris, R. P.; Purcell, Cormac; Rudnick, Lawrence; Sun, Xiaohui

doi:10.1093/mnras/sty2908

Cited by 14 publications

(9 citation statements)

References 26 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The first architecture equipped with the Inception module with dimension reduction was GoogLeNet, the winner of the ImageNet Large-Scale Visual Recognition Challenge 2014 (ILSVRC14; Russakovsky et al 2014). After the success of GoogLeNet, this module was also used in astronomical studies such as supernovae classification (Brunel et al 2019) and Faraday spectra classification (Brown et al 2019), and the mapping between simulation based galaxy cluster distributions and the underlying dark matter distribution (Zhang et al 2019). In this work, we explore its potential when creating Architecture G, see Table 4 and Section 5.5.…”

Section: Inception Modulementioning

confidence: 99%

Radio Galaxy Zoo: Giant Radio Galaxy Classification using Multi-Domain Deep Learning

Tang,

Scaife,

Wong

et al. 2021

Preprint

View full text Add to dashboard Cite

In this work we explore the potential of multi-domain multi-branch convolutional neural networks (CNNs) for identifying comparatively rare giant radio galaxies from large volumes of survey data, such as those expected for new generation radio telescopes like the SKA and its precursors. The approach presented here allows models to learn jointly from multiple survey inputs, in this case NVSS and FIRST, as well as incorporating numerical redshift information. We find that the inclusion of multi-resolution survey data results in correction of 39% of the misclassifications seen from equivalent single domain networks for the classification problem considered in this work. We also show that the inclusion of redshift information can moderately improve the classification of giant radio galaxies.

show abstract

Section: Inception Modulementioning

confidence: 99%

Radio Galaxy Zoo: Giant Radio Galaxy Classification using Multi-Domain Deep Learning

Tang,

Scaife,

Wong

et al. 2021

Preprint

View full text Add to dashboard Cite

show abstract

“…Stokes QU fitting [230,231], on the other hand, is a parametric approach to describe broadband polarization data using models of the magnetized medium along the line of sight. Recent developments in this area include the FIRESTARTER algorithm [238], which takes into account the spectral indices of each of the fitted polarized components, and the use of convolutional neural networks (CNN) to classify Faraday depth spectra, to distinguish simple sightlines that exhibit only one RM component from more complex sightlines [239]. Interpretation of Faraday spectra in the case of turbulence can be complicated [126] and will require additional consideration.…”

Section: Polarization-specific Processingmentioning

confidence: 99%

Magnetism Science with the Square Kilometre Array

Heald¹,

Mao

Vacca³

et al. 2020

Galaxies

Self Cite

View full text Add to dashboard Cite

The Square Kilometre Array (SKA) will answer fundamental questions about the origin, evolution, properties, and influence of magnetic fields throughout the Universe. Magnetic fields can illuminate and influence phenomena as diverse as star formation, galactic dynamics, fast radio bursts, active galactic nuclei, large-scale structure, and dark matter annihilation. Preparations for the SKA are swiftly continuing worldwide, and the community is making tremendous observational progress in the field of cosmic magnetism using data from a powerful international suite of SKA pathfinder and precursor telescopes. In this contribution, we revisit community plans for magnetism research using the SKA, in light of these recent rapid developments. We focus in particular on the impact that new radio telescope instrumentation is generating, thus advancing our understanding of key SKA magnetism science areas, as well as the new techniques that are required for processing and interpreting the data. We discuss these recent developments in the context of the ultimate scientific goals for the SKA era.

show abstract

“…This is the so called Faraday simple case (e.g. Brown et al 2019), where polarization angle, χ, is related to the Faraday depth by…”

Section: Rm Synthesismentioning

confidence: 99%

Broad-band radio polarimetry of disc galaxies and AGN with KAT-7

Legodi

Stil

2020

Monthly Notices of the Royal Astronomical Society

View full text Add to dashboard Cite

We report broad-band (1.2 - 1.9 GHz) radio continuum observations at arcminute resolutions of two nearby disk galaxies, NGC 1808 and NGC 1097, and four AGN powered radio sources; PKS B1934 − 638, PKS B0407 − 658, J0240 − 231, and J0538 − 440. We use Rotation Measure Synthesis to analyze their Faraday complexity. Observations were made with the KAT 7 radio telescope array, in South Africa. The AGN powered sources fall into two “Faraday” categories – simple and complex. The most polarized sources, J0538 − 440 and J0240 − 231, are found to have complex Faraday spectra that can be time variable (J0538 − 440 case) and also indicative of complex Faraday emitting and rotating components along the line of sight. PKS B0407 − 658 shows a simple Faraday spectrum while PKS B1934 − 638 is undetected in polarization. The disk galaxies are classified as complex, albeit at low signal-to-noise. This may indicate depolarization due to turbulence of the magnetised plasma in the bar and circumnuclear regions and/or frequency-dependent depolarization at L-band.

show abstract

Classifying Complex Faraday Spectra with Convolutional Neural Networks

Cited by 14 publications

References 26 publications

Radio Galaxy Zoo: Giant Radio Galaxy Classification using Multi-Domain Deep Learning

Radio Galaxy Zoo: Giant Radio Galaxy Classification using Multi-Domain Deep Learning

Magnetism Science with the Square Kilometre Array

Broad-band radio polarimetry of disc galaxies and AGN with KAT-7

Contact Info

Product

Resources

About