AGNs and quasars

As part of research on active galactic nuclei (AGN) and quasars (QSO), we develop automatic selection and redshift estimation methods of AGN and QSO in photometric surveys. Our main research interests include:
  • reliable and interpretable machine learning models
  • photo-z uncertainty
  • deep learning in low-dimensional spectroscopy and spectroscopic feature reconstruction from photometric data
  • model testing techniques
  • visualisation of high-dimensional feature spaces
  • extrapolation with predictions from spectroscopic surveys to much deeper photometric ones
  • type II quasar and AGN selection
  • outliers and unknown objects detection
  • problems with small training samples
  • SED fitting

Our efforts result in catalogs of AGN and QSO in popular surveys such as KiDS, SDSS, WISE and AKARI. We coordinate our work with the needs and goals of the LSST survey.