Estimating photometric redshifts using support vector machines

We present a new approach to obtaining photometric redshifts using a kernel learning technique called support vector machines. Unlike traditional spectral energy distribution fitting, this technique requires a large and representative training set. When one is available, however, it is likely to produce results that are comparable to the best results obtained using template fitting and artificial neural networks. Additional photometric parameters such as morphology, size, and surface brightness can be easily incorporated. The technique is demonstrated using samples of galaxies from the Sloan Digital Sky Survey Data Release 2 and the hybrid galaxy formation code GalICS. The rms error in redshift estimation is below 0.03 for both samples. The strengths and limitations of the technique are assessed.