Light curves neural network approximation

Most of the objects have never received a spectroscopic follow-up, which is especially crucial for transients, e.g., supernovae. Flux time series are actively used as an affordable alternative for photometric classification and characterization, e.g., peak identification and luminosity decline estimation.

However, the collected time series are multidimensional, irregularly sampled, may contain outliers, and do not have well-defined systematic uncertainties. This paper presents a search for the best-performing methods to approximate over time and wavelength the observed light curves for generating time series with regular time steps in each passband. We examine several light curve approximation methods based on neural networks such as Multilayer Perceptrons, Bayesian Neural Networks, and Normalizing Flows to approximate observations of a single light curve.