Support vector machines for photometric redshift measurement of quasars

Based on photometric and spectroscopic data of quasars from SDSS DR7 and UKDISS DR7, support vector machines (SVM) is applied to predict photometric redshifts of quasars. Different input patterns are tried and the best pattern is presented. Comparing the results using optical data with that using optical and infrared data, the experimental results show that the accuracy improves with data from more bands. In addition, the quasar sample is firstly clustered into two groups by one-class SVM, then the photometric redshifts of the two groups are separately estimated by means of SVM. The results based on the whole sample and the combined results from the two groups are comparable.