| Abstract: | We have undertaken a dedicated program of automatic source classification in the WISE database, comprehensively identifying galaxies, quasars and stars on most of the unconfused sky. We use the Support Vector Machines (SVM) classifier for that purpose, trained on SDSS spectroscopic data. We apply two photometric datasets: one is based on WISE information only, and employs its two shortest bands for optimal completeness and depth; the other adds all-sky photographic SuperCOSMOS B and R bands and provides a reliable sample of more local galaxies (zmed~0.2), as well as quasars. The results of our classification methods show very high purity and completness (more than 96%) of the separated sources (Kurcz et al., Krakowski et al. in prep) and the resultant catalogs might be used fo the sophisticaded analysis (e.g. all-sky photometric redshifts as presented by M. Bilicki in a separate talk). |