The Eb Factory Project. I. A Fast, Neural-Net-Based, General Purpose Light Curve Classifier Optimized for Eclipsing Binaries

Abstract

We describe a new neural-net based light curve classifier and provide it with documentation as a ready-to-use tool for the community. While optimized for identification and classification of eclipsing binary stars, the classifier is general purpose, and has been developed for speed in the context of upcoming massive surveys such as LSST. A challenge for classifiers in the context of neural-net training and massive data sets is to minimize the number of parameters required to describe each light curve. We show that a simple and fast geometric representation that encodes the overall light curve shape, together with a chi-square parameter to capture higherorder morphology information results in efficient yet robust light curve classification, especially for eclipsing binaries. Testing the classifier on the ASAS light curve database, we achieve a retrieval rate of 98% and a false-positive rate of 2% for eclipsing binaries. We achieve similarly high retrieval rates for most other periodic variable-star classes, including RR Lyrae, Mira, and delta Scuti. However, the classifier currently has difficulty discriminating between different subclasses of eclipsing binaries, and suffers a relatively low (∼60%) retrieval rate for multi-mode delta Cepheid stars. We find that it is imperative to train the classifier's neural network with exemplars that include the full range of light curve quality to which the classifier will be expected to perform; the classifier performs well on noisy light curves only when trained with noisy exemplars. The classifier source code, ancillary programs, a trained neural net, and a guide for use, are provided.

A machine learned classifier for RR Lyrae in the VVV survey

“…The performance of automated classifiers of variable sources in the optical has been assessed in several previous studies (e.g., Debosscher et al 2007;Dubath et al 2011;Richards et al 2011;Paegert et al 2014;Kim & Bailer-Jones 2016), which followed a similar approach. First we need to assess whether the variable is periodic, and if so, the period of the time series needs to be estimated.…”

The classification of periodic light curves from non-survey optimized observational data through automated extraction of phase-based visual features

“…This is very useful in astronomy due to the limitations in gathering data. In fact, this is one of the most powerful techniques in eliminating sampling issues as long as the light curve does have a dominant period [7]. For non-periodic variable objects in astronomy, such as transient light sources, other approaches must be considered.…”

“…Other important variable light sources are eclipsing binary systems. These systems exhibit periodic brightness changes from binary stars eclipsing each other due to the orbital plane of the system having a low inclination relative to the Earth [7].…”

“…Puegert et al have also proposed the extraction of shape-defining features from phase-folded light curves through the coefficients of fitted chains of polynomial models. These features were then used to classify eclipsing binary light curves using learned models built using a feedforward neural network [7].…”

Classifying Periodic Astrophysical Phenomena from non-survey optimized variable-cadence observational data

“…The original PolyFit implementation placed the four knots where the light curve data points crossed the mean magnitude of the light curve, randomly perturbed the knots using a random Gaussian 'kick' and then allowed them to relax into a minimum χ 2 state over a small number of iterations (Paegert et al, 2014). Each iteration must be carefully checked as the phase intervals must have an appropriate number of data points to prevent degeneracy in the polynomial fits.…”