On the cosmological performance of photometric classified supernovae with machine learning

The efficient classification of the different types of Supernova is one of the most important problems of the observational cosmology. Spectroscopic confirmation of most objects in upcoming photometric surveys, such as the Dark Energy Survey and the Large Synoptic Survey Telescope, will be nevertheless impossible. The development of automated classification process based on photometry has thus become a great challenge. In this paper we investigate the performance of machine learning classification on the final cosmological constraints using simulated lightcurves from The Supernova Photometric Classification Challenge, released in 2010. We study the use of different feature sets for the lightcurves and many different ML pipelines based on either decision tree ensembles or automated search processes. To construct the final catalogs we propose a threshold selection method, by employing a Bias-Variance tradeoff. This is a very robust and efficient way to minimize the Mean Squared Error. With it we were able to get very strong cosmological performance, which was able to keep ~ 75% of the total information in the type Ia SNe in the case of the SALT2 feature set and ~33% in the case of other feature sets (based on either the Newling model or on standard wavelet decomposition)