Evaluating the optical classification of Fermi BCUs using machine learning. (arXiv:1902.07717v1 [astro-ph.HE])
<a href="http://arxiv.org/find/astro-ph/1/au:+Kang_S/0/1/0/all/0/1">Shi-Ju Kang</a>, <a href="http://arxiv.org/find/astro-ph/1/au:+Fan_J/0/1/0/all/0/1">Junhui Fan</a>, <a href="http://arxiv.org/find/astro-ph/1/au:+Mao_W/0/1/0/all/0/1">Weiming Mao</a>, <a href="http://arxiv.org/find/astro-ph/1/au:+Wu_Q/0/1/0/all/0/1">Qingwen Wu</a>, <a href="http://arxiv.org/find/astro-ph/1/au:+Feng_J/0/1/0/all/0/1">Jianchao Feng</a>, <a href="http://arxiv.org/find/astro-ph/1/au:+Yin_Y/0/1/0/all/0/1">Yue Yin</a>
In the third catalog of active galactic nuclei detected by the Fermi-LAT
(3LAC) Clean Sample, there are 402 blazars candidates of uncertain type (BCU).
Due to the limitations of astronomical observation or intrinsic properties, it
is difficult to classify blazars using optical spectroscopy. The potential
classification of BCUs using machine learning algorithms is essential. Based on
the 3LAC Clean Sample, we collect 1420 Fermi blazars with 8 parameters of
{gamma}-ray photon spectral index, radio flux, flux density, curve
significance, the integral photon flux in 100 to 300 MeV, 0.3 to 1 GeV, 10 to
100 GeV and variability index. Here, we apply 4 different supervised machine
learning (SML) algorithms (emph{Decision trees, Random forests, support vector
machines and Mclust Gaussian finite mixture models}) to evaluate the
classification of BCUs based on the direct observational properties. All the 4
methods can perform exceedingly well with a more accuracy and can effective
forecast the classification of Fermi BCUs. The evaluating results show the
results of these methods (SML) are valid and robust, where, about 1/4 sources
are FSRQs and 3/4 are BL Lacs in 400 BCUs, which are consistent with some other
recent results. Although a number of factors influence the accuracy of SML, the
results are stable at a fixed ratio 1:3 between FSRQs and BL Lacs, which
suggests that the SML can provides an effective method to evaluate the
potential classification of BCUs. Among the 4 methods, Mclust Gaussian Mixture
Modelling has the highest accuracy for our training sample (4/5, seed=123).
In the third catalog of active galactic nuclei detected by the Fermi-LAT
(3LAC) Clean Sample, there are 402 blazars candidates of uncertain type (BCU).
Due to the limitations of astronomical observation or intrinsic properties, it
is difficult to classify blazars using optical spectroscopy. The potential
classification of BCUs using machine learning algorithms is essential. Based on
the 3LAC Clean Sample, we collect 1420 Fermi blazars with 8 parameters of
{gamma}-ray photon spectral index, radio flux, flux density, curve
significance, the integral photon flux in 100 to 300 MeV, 0.3 to 1 GeV, 10 to
100 GeV and variability index. Here, we apply 4 different supervised machine
learning (SML) algorithms (emph{Decision trees, Random forests, support vector
machines and Mclust Gaussian finite mixture models}) to evaluate the
classification of BCUs based on the direct observational properties. All the 4
methods can perform exceedingly well with a more accuracy and can effective
forecast the classification of Fermi BCUs. The evaluating results show the
results of these methods (SML) are valid and robust, where, about 1/4 sources
are FSRQs and 3/4 are BL Lacs in 400 BCUs, which are consistent with some other
recent results. Although a number of factors influence the accuracy of SML, the
results are stable at a fixed ratio 1:3 between FSRQs and BL Lacs, which
suggests that the SML can provides an effective method to evaluate the
potential classification of BCUs. Among the 4 methods, Mclust Gaussian Mixture
Modelling has the highest accuracy for our training sample (4/5, seed=123).
http://arxiv.org/icons/sfx.gif
