Pan-STARRS1 Source Types and Redshifts with Machine Learning ("PS1-STRM")

PS1-STRM is a neural network source classification and photometric redshift catalog created from the PanSTARRS1 (PS1) 3π DR1. Neural networks have been trained on a compilation of spectroscopic measurements, cross-matched with PS1. Based on PS1 forced mean photometry, a source is classified as galaxy, star, quasar, or unsure. For galaxies, photometric redshift estimation is performed, also yielding an estimate of redshift error via Monte-Carlo sampling. Sources lying outside the parameter coverage of the training set (i.e. extrapolated sources) are identified using self-organizing maps. Classification and photo-z results are provided for every source in the PS1 3π DR1 ForcedMeanObject table, a total of 2,902,054,648 objects. See the README file and the primary reference paper for a detailed description of the catalog metadata.