Outlier Detection in the DESI Bright Galaxy Survey

TL;DR

This paper introduces an unsupervised method using autoencoders and normalizing flows to identify rare and unusual objects in the DESI Bright Galaxy Survey, revealing new astrophysical phenomena and misclassifications.

Contribution

The study develops a novel unsupervised outlier detection approach combining autoencoders and normalizing flows for galaxy spectra analysis, and provides a publicly available catalog and models.

Findings

Identified diverse astrophysical outliers including rare quasars and galaxy mergers.

Discovered that many outliers are stars misclassified as galaxies due to PCA limitations.

Provided a comprehensive catalog and models for future follow-up studies.

Abstract

We present an unsupervised search for outliers in the Bright Galaxy Survey (BGS) dataset from the DESI Early Data Release. This analysis utilizes an autoencoder to compress galaxy spectra into a compact, redshift-invariant latent space, and a normalizing flow to identify low-probability objects. The most prominent outliers show distinctive spectral features such as irregular or double-peaked emission lines, or originate from galaxy mergers, blended sources, and rare quasar types, including one previously unknown Broad Absorption Line system. A significant portion of the BGS outliers are stars spectroscopically misclassified as galaxies. By building our own star model trained on spectra from the DESI Milky Way Survey, we have determined that the misclassification likely stems from the Principle Component Analysis of stars in the DESI pipeline. To aid follow-up studies, we make the full probability catalog of all BGS objects and our pre-trained models publicly available.