Using Neural Networks to Automate the Identification of Brightest Cluster Galaxies in Large Surveys

TL;DR

This paper introduces a neural network-based method to automatically identify Brightest Cluster Galaxies in large surveys using multiband photometric images, demonstrating high accuracy especially with representative data.

Contribution

The authors develop and validate a neural network approach for BCG identification that works with minimal information, improving efficiency in large survey analyses.

Findings

Achieves R² ≈ 0.94 on simulated data

Achieves R² ≈ 0.60 on real data, improving to 0.99 with representative samples

Performs reliably up to redshift z ≈ 0.6

Abstract

Brightest cluster galaxies (BCGs) lie deep within the largest gravitationally bound structures in existence. Though some cluster finding techniques identify the position of the BCG and use it as the cluster center, other techniques may not automatically include these coordinates. This can make studying BCGs in such surveys difficult, forcing researchers to either adopt oversimplified algorithms or perform cumbersome visual identification. For large surveys, there is a need for a fast and reliable way of obtaining BCG coordinates. We propose machine learning to accomplish this task and train a neural network to identify positions of candidate BCGs given no more information than multiband photometric images. We use both mock observations from The Three Hundred project and real ones from the Sloan Digital Sky Survey (SDSS), and we quantify the performance. Training on simulations yields a squared correlation coefficient, R$^2$, between predictions and ground truth of R$^2 \approx 0.94$ when testing on simulations, which decreases to R$^2 \approx 0.60$ when testing on real data due to discrepancies between datasets. Limiting the application of this method to real clusters more representative of the training data, such those with a BCG r-band magnitude $r_{\text{BCG}} \leq 16.5$, yields R$^2 \approx 0.99$. The method performs well up to a redshift of at least $z\approx 0.6$. We find this technique to be a promising method to automate and accelerate the identification of BCGs in large datasets.