Simone Mender

Search for Sources of Rare Types by Cluster Analysis(pdf)

The AGN unification model includes different extragalactic source types. Their characteristics are investigated using multiwavelength astronomy. Machine learning with these multiwavelength data is one opportunity to classify the sources. Supervised machine learning methods need labeled data for training. In the search of rare types, such as dust obscured blazars, methods requiring labeled data are not suitable, because only a very few such sources are known. In contrast cluster analysis - a type of unsupervised machine learning - is able to uncover structures of unlabeled data. This talk gives an introduction into cluster analysis and its application in astroparticle physics. The idea of searching for rare source candidates with this method will be presented.