Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Towards a foundation model for astrophysical source detection: An End-to-End Gamma-Ray Data Analysis Pipeline Using Deep Learning

Judit Pérez-Romero, Saptashwa Bhattacharyya, Sascha Caron, Dmitry Malyshev, Rodney Nicolas, Giacomo Principe, Zoja Rokavec, Roberto Ruiz de Austri, Danijel Skočaj, Fiorenzo Stoppa, Domen Tabernik, Gabrijela Zaharijas

TL;DR

This work extends the AutoSourceID (ASID) method, initially tested with Fermi-LAT simulated data and optical data and optical data, to Cherenkov Telescope Array Observatory (CTAO) simulated data, and demonstrates a versatile framework for future application to other surveys.

Abstract

The increasing volume of gamma-ray data demands new analysis approaches that can handle large-scale datasets while providing robustness for source detection. We present a Deep Learning (DL) based pipeline for detection, localization, and characterization of gamma-ray sources. We extend our AutoSourceID (ASID) method, initially tested with \textit{Fermi}-LAT simulated data and optical data (MeerLICHT), to Cherenkov Telescope Array Observatory (CTAO) simulated data. This end-to-end pipeline demonstrates a versatile framework for future application to other surveys and potentially serves as a building block for a foundational model for astrophysical source detection.

Towards a foundation model for astrophysical source detection: An End-to-End Gamma-Ray Data Analysis Pipeline Using Deep Learning

TL;DR

This work extends the AutoSourceID (ASID) method, initially tested with Fermi-LAT simulated data and optical data and optical data, to Cherenkov Telescope Array Observatory (CTAO) simulated data, and demonstrates a versatile framework for future application to other surveys.

Abstract

The increasing volume of gamma-ray data demands new analysis approaches that can handle large-scale datasets while providing robustness for source detection. We present a Deep Learning (DL) based pipeline for detection, localization, and characterization of gamma-ray sources. We extend our AutoSourceID (ASID) method, initially tested with \textit{Fermi}-LAT simulated data and optical data (MeerLICHT), to Cherenkov Telescope Array Observatory (CTAO) simulated data. This end-to-end pipeline demonstrates a versatile framework for future application to other surveys and potentially serves as a building block for a foundational model for astrophysical source detection.

Paper Structure

This paper contains 5 sections, 1 equation, 3 figures.

Table of Contents

  1. Introduction
  2. Fermi-LAT: towards the first DL gamma-ray catalog
  3. CTAO: challenges of the new generation of IACTS
  4. Towards a foundation model: optical data and the latent space
  5. Conclusions

Figures (3)

  • Figure 1: Workflow of the updated ASID pipeline consisting in the source detection and localization module, followed by the modules on source classification, position refinement, and flux estimation including uncertainties.
  • Figure 2: Example patches of ASID (left panel) and CeDIRNet (right panel) results using log-scaling on the image. The patches correspond to different realizations. The white stars represent true sources and the green circles the detected sources, and the green crosses the fake positives. The colormaps represent the number of counts.
  • Figure 3: Right: Example patch with localized sources by ASID (red circles) superimposed on the optical image in the presence of artifacts ( Stoppa:2022wta). Left: Representation of the latent space of the bottleneck of our model. Grey (Fermi-LAT) and brown (CTAO) points represent the gamma-ray backgrounds, while red (Fermi-LAT) and pink (CTAO) represent the gamma-ray sources.