Browsing by Author "Taylor, A R"

Now showing 1 - 2 of 2
  • Thumbnail Image
    Item
    Open Access
    Compact and Extended Radio Sources Classification using Deep Convolutional Neural Networks
    (2019) Alhassan, Wathela; Taylor, A R; Vaccari, Mattia
    Upcoming surveys with new radio observatories such as the Square Kilometer Array will generate a wealth of imaging data containing large numbers of radio sources. Different classes of radio sources can be used as tracers of the cosmic environment, including the dark matter density field, to address key cosmological questions. Classifying these sources based on morphology is thus an important step toward achieving the science goals of next generation radio surveys. Extended Radio Sources have been traditionally classified as Fanaroff-Riley (FR) I and II, although some exhibit more complex 'bent' morphologies arising from environmental factors or intrinsic properties. In this work we present the FIRST Classifier, an on-line system for automated classification of Compact and Extended radio sources. We developed the FIRST Classifier based on a trained Deep Convolutional Neural Network Model to automate the morphological classification of compact and extended radio sources observed in the FIRST radio survey. Our model was trained independently for 20 times and achieved an average accuracy, precision, recall and F1 of 0.98. The current version of the FIRST classifier is able to identify the morphological class for a single source or for a list of sources as Compact or Extended (FRI, FRII and BENT).
  • Thumbnail Image
    Item
    Open Access
    DEGREE OF POLARIZATION AND SOURCE COUNTS OF FAINT RADIO SOURCES FROM STACKING POLARIZED INTENSITY
    (2014) Stil, J M; Keller, B W; George, S J; Taylor, A R
    We present stacking polarized intensity as a means to study the polarization of sources that are too faint to be detected individually in surveys of polarized radio sources. Stacking offers not only high sensitivity to the median signal of a class of radio sources, but also avoids a detection threshold in polarized intensity, and therefore an arbitrary exclusion of source with a low percentage of polarization. Correction for polarization bias is done through a Monte Carlo analysis and tested on a simulated survey. We show that the non-linear relation between the real polarized signal and the detected signal requires knowledge of the shape of the distribution of fractional polarization, which we constrain using the ratio of the upper quartile to the lower quartile of the distribution of stacked polarized intensities. Stacking polarized intensity for NVSS sources down to the detection limit in Stokes I, we find a gradual increase in median fractional polarization that is consistent with a trend that was noticed before for bright NVSS sources, but is much more gradual than found by previous deep surveys of radio polarization. Consequently, the polarized radio source counts derived from our stacking experiment predict fewer polarized radio sources for future surveys with the Square Kilometre Array and its pathfinders.