A simple and robust method for automated photometric classification of supernovae using neural networks

TL;DR

This paper introduces a neural network-based method for automated photometric classification of supernovae, achieving high accuracy without requiring spectroscopic data, and can be extended for detailed supernova subclassification.

Contribution

The authors develop a two-step neural network approach that uses fitted lightcurve parameters and uncertainties for supernova classification, demonstrating improved performance and flexibility over previous methods.

Findings

Achieved 78-82% completeness and purity in supernova classification.

Including host-galaxy redshift data modestly improves accuracy.

Classification quality remains stable across different redshifts.

Abstract

A method is presented for automated photometric classification of supernovae (SNe) as Type-Ia or non-Ia. A two-step approach is adopted in which: (i) the SN lightcurve flux measurements in each observing filter are fitted separately; and (ii) the fitted function parameters and their associated uncertainties, along with the number of flux measurements, the maximum-likelihood value of the fit and Bayesian evidence for the model, are used as the input feature vector to a classification neural network (NN) that outputs the probability that the SN under consideration is of Type-Ia. The method is trained and tested using data released following the SuperNova Photometric Classification Challenge (SNPCC). We consider several random divisions of the data into training and testing sets: for instance, for our sample D_1 (D_4), a total of 10% (40%) of the data are involved in training the algorithm and the remainder used for blind testing of the resulting classifier; we make no selection cuts. Assigning a canonical threshold probability of p_th=0.5 on the NN output to classify a SN as Type-Ia, for the sample D_1 (D_4) we obtain a completeness of 0.78 (0.82), purity of 0.77 (0.82), and SNPCC figure-of-merit of 0.41 (0.50). Including the SN host-galaxy redshift and its uncertainty as additional inputs to the NN results in a modest 5-10% increase in these values. We find that the classification quality does not vary significantly with SN redshift. Moreover, our probabilistic classification method allows one to calculate the expected completeness, purity or other measures of classification quality as a function of the p_th, without knowing the true classes of the SNe in the testing sample. The method may thus be improved further by optimising p_th and can easily be extended to divide non-Ia SNe into their different classes.