Machine Learning Search for Gamma-Ray Burst Afterglows in Optical Surveys

TL;DR

This paper develops machine learning methods to automatically identify gamma-ray burst afterglows in optical survey data, achieving over 90% accuracy, aiding rapid follow-up and statistical analysis of transient events.

Contribution

It introduces a combined meta-classifier that outperforms individual models in classifying GRB afterglows using optical color indices.

Findings

Meta-classifier achieves >90% accuracy in identifying GRB afterglows.

Support vector machine, random forest, and neural network are compared.

Optical color indices effectively distinguish GRB afterglows from other objects.

Abstract

Thanks to the advances in robotic telescopes, the time domain astronomy leads to a large number of transient events detected in images every night. Data mining and machine learning tools used for object classification are presented. The goal is to automatically classify transient events for both further follow-up by a larger telescope and for statistical studies of transient events. A special attention is given to the identification of gamma-ray burst afterglows. Machine learning techniques is used to identify GROND gamma-ray burst afterglow among the astrophysical objects present in the SDSS archival images based on the $g'-r'$, $r'-i'$ and $i'-z'$ colour indices. The performance of the support vector machine, random forest and neural network algorithms is compared. A joint meta-classifier, built on top of the individual classifiers, can identify GRB afterglows with the overall accuracy of $\gtrsim 90\%$.