Using cGANs for Anomaly Detection: Identifying Astronomical Anomalies in JWST NIRcam Imaging

TL;DR

This paper demonstrates that cGANs can effectively predict certain galaxy fluxes in JWST data and identify anomalies by failing on rare objects, offering a new approach for astronomical anomaly detection.

Contribution

The study introduces a cGAN-based method for anomaly detection in JWST imaging data, highlighting its ability to identify rare astronomical objects by prediction failures.

Findings

cGAN accurately predicts fluxes of known galaxy types

Fails to predict fluxes for rare objects like galaxy mergers

Potential for anomaly detection in astronomical surveys

Abstract

We present a proof of concept for mining JWST imaging data for anomalous galaxy populations using a conditional Generative Adversarial Network (cGAN). We train our model to predict long wavelength NIRcam fluxes (LW: F277W, F356W, F444W between 2.4 to 5.0\mu m) from short wavelength fluxes (SW: F115W, F150W, F200W between 0.6 to 2.3\mu m) in approximately 2000 galaxies. We test the cGAN on a population of 37 Extremely Red Objects (EROs) discovered by the CEERS JWST Team arXiv:2305.14418. Despite their red long wavelength colours, the EROs have blue short wavelength colours (F150W \- F200W equivalently 0 mag) indicative of bimodal SEDs. Surprisingly, given their unusual SEDs, we find that the cGAN accurately predicts the LW NIRcam fluxes of the EROs. However, it fails to predict LW fluxes for other rare astronomical objects, such as a merger between two galaxies, suggesting that the cGAN can be used to detect some anomalies