Extracting latent representations from X-ray spectra. Classification, regression, and accretion signatures of Chandra sources

Abstract

Spectral signatures are crucial in the era of large X-ray surveys. Automatic machine learning methods have proven useful in this respect, but so far they have not been applied to large spectral datasets, such as the Chandra Source Catalog (CSC). This work aims to develop a compact and physically meaningful representation of Chandra X-ray spectra using deep learning. To verify that the learned representation captures relevant information, we evaluate it through classification, regression, and interpretability analyses, and measure the mutual information between spectral and time-domain properties of these sources, aiding in the future identification of transient events. We use a transformer-based autoencoder to compress X-ray spectra into representations in an 8-dimensional latent space. Astrophysical source types and physical summary statistics are compiled from external catalogs. We evaluate the learned representation in terms of spectral reconstruction accuracy, clustering performance on 8 known astrophysical source classes, and correlation with physical quantities such as hardness ratios and hydrogen column densities ($N_H$). Upon reconstruction, clustering in the latent space yields a balanced classification accuracy of $\sim$40% across the 8 source classes, increasing to $\sim$69% when restricted to AGNs and stellar-mass compact objects exclusively. Moreover, latent features correlate with spectral and temporal properties, suggesting that the compressed representation captures physically relevant information. Features learned directly from X-ray spectra capture relevant physical information as effectively as human-extracted features that require additional computations. They can be used for both classification and regression in large surveys, and also share mutual information with time-domain properties. The method can be adapted to existing and upcoming X-ray catalogs.

21 pages, 17 figures; accepted in A&A

References in corpus (31)

• Collisional Plasma Models with APEC/APED: Emission Line Diagnostics of Hydrogen-like and Helium-like Ions
• The eROSITA X-ray telescope on SRG
• The Chandra Deep Field-South Survey: 7 Ms Source Catalogs
• The XMM-Newton serendipitous survey IX. The fourth XMM-Newton serendipitous source catalogue
• Interpretable Scientific Discovery with Symbolic Regression: A Review
• Nine-hour X-ray quasi-periodic eruptions from a low-mass black hole galactic nucleus
• X-ray Quasi-Periodic Eruptions from two previously quiescent galaxies
• 2SXPS: An improved and expanded Swift X-ray telescope point source catalog
• Relational Autoencoder for Feature Extraction
• Analysis of Agglomerative Clustering
• Self-supervised Learning for Human Activity Recognition Using 700,000 Person-days of Wearable Data
• The NewAthena mission concept in the context of the next decade of X-ray astronomy
• Identifying Strong Lenses with Unsupervised Machine Learning using Convolutional Autoencoder
• Beyond the Hubble Sequence -- Exploring Galaxy Morphology with Unsupervised Machine Learning
• AstroCLIP: A Cross-Modal Foundation Model for Galaxies
• Extragalactic fast X-ray transient candidates discovered by Chandra (2000-2014)
• Deblending galaxies with Variational Autoencoders: a joint multi-band, multi-instrument approach
• A review of unsupervised learning in astronomy
• Anomaly detection in Hyper Suprime-Cam galaxy images with generative adversarial networks
• Discovery of Three Candidate Magnetar-powered Fast X-ray Transients from Chandra Archival Data
• Classifying Unidentified X-ray Sources in the Chandra Source Catalog Using a Multiwavelength Machine-learning Approach
• An autoencoder of stellar spectra and its application in automatically estimating atmospheric parameters
• Empirical Links between XRB and AGN accretion using the complete z<0.4 spectroscopic CSC/SDSS Catalog
• Deducing Neutron Star Equation of State Parameters Directly From Telescope Spectra with Uncertainty-Aware Machine Learning
• Real-Time Detection of Anomalies in Large-Scale Transient Surveys
• Autoencoding Galaxy Spectra II: Redshift Invariance and Outlier Detection
• Unsupervised Machine Learning for the Classification of Astrophysical X-ray Sources
• Representation Learning for Time-Domain High-Energy Astrophysics: Discovery of Extragalactic Fast X-ray Transient XRT 200515
• DeepGraviLens: a Multi-Modal Architecture for Classifying Gravitational Lensing Data
• Rapid Spectral Parameter Prediction for Black Hole X-Ray Binaries using Physicalised Autoencoders
• Deciphering galaxy images using machine vision -- Combining variational autoencoder and principal component analysis for feature extraction