Presentation #107.39 in the session “ISM/Galaxies/Clusters (Poster)”.
The same relativistic electrons that produce synchrotron radio emission in galaxy clusters are expected to produce hard X-ray emission through inverse-Compton (IC) scattering of the cosmic microwave background. The measurements of IC emission would enhance our understanding of the non-thermal processes as well as the magnetic fields of the intra-cluster medium. To date, no clear detection of the emission has been made in galaxy clusters despite the dedicated searching with NuSTAR. This is mostly due to its flux being orders of magnitude smaller than the underlying thermal emission. In this work, we apply a conditional auto-encoder to detect the IC emission from a given X-ray spectrum. This machine learning model is trained solely on synthetic NuSTAR spectra of thermal emission with realistic background noises included. We show that without the prior knowledge of IC emission, the auto-encoder is capable of detecting the additional IC emission. This semi-supervised machine learning approach can aid the detection of the long sought IC emission and help calibrate the non-thermal phenomena for cluster cosmology.