Photometric selection of quasars in large astronomical data sets with a fast and accurate machine learning algorithm

Future astronomical surveys will produce data on ∼108 objects per night. In order to characterize and classify these sources, we will require algorithms that scale linearly with the size of the data, that can be easily parallelized and where the speedup of the parallel algorithm will be linear in the number of processing cores. In this paper, we present such an algorithm and apply it to the questi...