Poster
2025 workshop program
Expediting Black Hole X-Ray Spectroscopy: Variational Auto-Encoders with Normalizing Flows
The work
Abstract
Black hole X-ray binaries (BHBs) can be studied with spectral fitting to provide physical constraints on accretion in extreme gravitational environments. We employ thousands of spectra collected by NICER, applying machine learning methods to expedite spectral-fitting results. Specifically, we create a probabilistic model, utilizing a variational auto-encoder with a normalizing flow prior, trained to adopt a physical latent space, building on our earlier work. This neural network produces predictions for spectral-model parameters as well as their full probability distributions.