Measuring redshift and galaxy properties via a multi-task neural net with probabilistic outputs: An application to simulated MOONS spectra

Kavli Affiliate: Roberto Maiolino

| First 5 Authors: Michele Ginolfi, Filippo Mannucci, Francesco Belfiore, Alessandro Marconi, Nicholas Boardman

| Summary:

The era of large-scale astronomical surveys demands innovative approaches for
rapid and accurate analysis of extensive spectral data, and a promising
direction to address this challenge is offered by artificial intelligence (AI).
Here we introduce a new pipeline, M-TOPnet (Multi-Task network Outputting
Probabilities), which employs a convolutional neural network (CNN) with
residual learning to simultaneously derive redshift and other key physical
properties of galaxies from their spectra. Our tool efficiently encodes
spectral information into a latent space, employing distinct downstream
branches for each physical quantity, thereby benefiting from multi-task
learning. Notably, our method handles the redshift output as a probability
distribution, allowing for a more refined and robust estimation of this
critical parameter. We demonstrate preliminary results using simulated data
from the MOONS instrument, which will be soon operating at the ESO/VLT. We
highlight the effectiveness of our tool in accurately predicting redshift,
stellar mass, and star-formation rate for galaxies at z>~1-3, even for faint
sources (m_H >~ 24) where traditional methods often struggle. Through analysis
of the output probability distributions, we demonstrate that our pipeline
enables robust quality screening of the results, achieving accuracy rates of up
to 99% in redshift determination (defined as predictions within |Delta_z| <
0.01 relative to the true redshift) with 8 h exposure spectra, while
automatically identifying potentially problematic cases. Our AI pipeline thus
emerges as a powerful solution for the upcoming challenges in observational
astronomy, combining precision, interpretability, and efficiency, all aspects
which are crucial for analysing the massive datasets expected from
next-generation instruments.

| Search Query: ArXiv Query: search_query=au:”Roberto Maiolino”&id_list=&start=0&max_results=3

Read More