<p>The dataset used to select active galactic nucleus (AGN) candidates is labelled as <strong>G_all.csv.</strong> The two datasets selected by the two ensembles of neural controllers are labelled as <strong>G_0_ensemble.csv</strong> and <strong>G_all_ensemble.csv</strong> respectively. The dataset of detected AGN is labelled as <strong>agn.csv.</strong> In these four datasets, each source is described by the following columns:</p>
<ul>
<li>ra (right ascension)</li>
<li>dec (declination)</li>
<li>z (photmetric redshift)</li>
<li>sfr (star-formation rate [log solar mass per year])</li>
<li>mass (stellar mass [log solar mass])</li>
<li>ks (Ks aperture magnitude)</li>
</ul>
<p>The final four columns are the IRAC filters, with the values following the underscore indicating the band (in units of micrometer):</p>
<ul>
<li>S_3.6</li>
<li>S_4.5</li>
<li>S_5.8</li>
<li>S_8.0</li>
</ul>
<p>The values for these fields are extracted from the COSMOS2015 multiwavelength catalogue. A non-detection in any of the fields is indicated by the 'NA' keyword. </p>
<p>The .fits files for the four radiomaps are also attached:</p>
<ul>
<li>VLA 3GHz</li>
<li>VLA 3GHz smoothed</li>
<li>VLA 1.4 GHz</li>
<li>MeerKAT</li>
</ul>