A Comparative Study of Machine Learning Methods for X-ray Binary Classification

TL;DR

This study compares three machine learning methods—Bayesian Gaussian Processes, K-Nearest Neighbors, and Support Vector Machines—for classifying X-ray binary systems into neutron stars or black holes using 3D color-color-intensity diagrams derived from six years of MAXI/GSC data, achieving high accuracy but still facing some confusion between subclasses.

Contribution

It introduces a comparative analysis of ML classification methods for XRBs using spatial patterns in 3D diagrams, highlighting their effectiveness and limitations.

Findings

All methods accurately distinguish pulsars from NPNS (95-100%).

High accuracy (92%) in differentiating BHs from pulsars.

KNN best predicts BHs among the tested methods.

Abstract

X-ray Binaries (XRBs) consist of a compact object that accretes material from an orbiting secondary star. The most secure method we have for determining if the compact object is a black hole is to determine its mass: this is limited to bright objects, and requires substantial time-intensive spectroscopic monitoring. With new X-ray sources being discovered with different X-ray observatories, developing efficient, robust means to classify compact objects becomes increasingly important. We compare three machine learning classification methods (Bayesian Gaussian Processes (BGP), K-Nearest Neighbors (KNN), Support Vector Machines (SVM)) for determining the compact objects as neutron stars or black holes (BHs) in XRB systems. Each machine learning method uses spatial patterns which exist between systems of the same type in 3D Color-Color-Intensity diagrams. We used lightcurves extracted using six years of data with MAXI/GSC for 44 representative sources. We find that all three methods are highly accurate in distinguishing pulsing from non-pulsing neutron stars (NPNS) with 95\% of NPNS and 100\% of pulsars accurately predicted. All three methods have high accuracy distinguishing BHs from pulsars (92\%) but continue to confuse BHs with a subclass of NPNS, called the Bursters, with KNN doing the best at only 50\% accuracy for predicting BHs. The precision of all three methods is high, providing equivalent results over 5-10 independent runs. In a future work, we suggest a fourth dimension be incorporated to mitigate the confusion of BHs with Bursters. This work paves the way towards more robust methods to efficiently distinguish BHs, NPNS, and pulsars.