A refined QSO selection method using diagnostics

We present 663 QSO candidates in the Large Magellanic Cloud (LMC) that were selected using multiple diagnostics. We started with a set of 2,566 QSO candidates selected using the methodology presented in our previous work based on time variability of the MACHO LMC light curves. We then obtained additional information for the candidates by cross-matching them with the Spitzer SAGE, the 2MASS, the Chandra, the XMM, and an LMC UBVI catalogues. Using that information, we specified diagnostic features based on mid-IR colours, photometric redshifts using SED template fitting, and X-ray luminosities, in order to discriminate more high-confidence QSO candidates in the absence of spectral information. We then trained a one-class Support Vector Machine model using those diagnostics features. We applied the trained model to the original candidates, and finally selected 663 high-confidence QSO candidates. We cross-matched those 663 QSO candidates with 152 newly-confirmed QSOs and 275 non-QSOs in the LMC fields, and found that the false positive rate was less than 1%.