arXiv Open Access 2019

Hybrid Physical-Deep Learning Model for Astronomical Inverse Problems

Francois Lanusse Peter Melchior Fred Moolekamp

Lihat Sumber

Abstrak

We present a Bayesian machine learning architecture that combines a physically motivated parametrization and an analytic error model for the likelihood with a deep generative model providing a powerful data-driven prior for complex signals. This combination yields an interpretable and differentiable generative model, allows the incorporation of prior knowledge, and can be utilized for observations with different data quality without having to retrain the deep network. We demonstrate our approach with an example of astronomical source separation in current imaging data, yielding a physical and interpretable model of astronomical scenes.

Topik & Kata Kunci

astro-ph.IM cs.LG

Penulis (3)

Francois Lanusse

Peter Melchior

Fred Moolekamp

Format Sitasi

APA MLA BibTeX

Lanusse, F., Melchior, P., Moolekamp, F. (2019). Hybrid Physical-Deep Learning Model for Astronomical Inverse Problems. https://arxiv.org/abs/1912.03980

Akses Cepat

Lihat di Sumber

Informasi Jurnal

Tahun Terbit: 2019
Bahasa: en
Sumber Database: arXiv
Akses: Open Access ✓