Convolutional Neural Networks for classifying galaxy mergers: Can faint tidal features aid in classifying mergers?

TL;DR

The paper investigates whether faint tidal features in LSST-depth images can enhance CNN-based merger classification using high-resolution TNG50 simulations to generate ground-truth labels. A suite of LSST-like surface-brightness maps is created for 151 Milky Way–like centrals, with extensive data augmentation and a modified CNN architecture inspired by prior work. Hyperparameter tuning and image-processing variants reveal that incorporating faint features yields a modest but meaningful improvement, achieving about 67–70% accuracy, compared with 65–67% for the baseline. The findings suggest faint tidal features carry usable signal for merger identification but underscore the need for larger, more diverse datasets and refined labeling to reach robust performance in real surveys.

Abstract

Identifying mergers from observational data has been a crucial aspect of studying galaxy evolution and formation. Tidal features, typically fainter than 26 ${\rm mag\,arcsec^{-2}}$, exhibit a diverse range of appearances depending on the merger characteristics and are expected to be investigated in greater detail with the Rubin Observatory Large Synoptic Survey Telescope (LSST), which will reveal the low surface brightness universe with unprecedented precision. Our goal is to assess the feasibility of developing a convolutional neural network (CNN) that can distinguish between mergers and non-mergers based on LSST-like deep images. To this end, we used Illustris TNG50, one of the highest-resolution cosmological hydrodynamic simulations to date, allowing us to generate LSST-like mock images with a depth $\sim$ 29 ${\rm mag\,arcsec^{-2}}$ for low-redshift ($z=0.16$) galaxies, with labeling based on their merger status as ground truth. We focused on 151 Milky Way-like galaxies in field environments, comprising 81 non-mergers and 70 mergers. After applying data augmentation and hyperparameter tuning, a CNN model was developed with an accuracy of 65--67\%. Through additional image processing, the model was further optimized, achieving an accuracy of 67--70\% when trained on images containing only faint features. This represents an improvement of $\sim$ 5\% compared to training on images with bright features only. This suggests that faint tidal features can serve as effective indicators for distinguishing between mergers and non-mergers. The future direction for further improvement based on this study is also discussed.

Figures (4)

• Figure 1: The top and bottom rows display example galaxies with and without tidal features, respectively. Each column represents different image processing methods: (a) original images with no mask (NM), (b) images after masking faint features (MF), (c) images after masking bright features (MB), and (d) images after masking bright features and inverting unmasked, star-particle pixels (MBI).
• Figure 2: Training curves for accuracy and loss of Fiducial3 (top) and Fiducial28 (bottom). Crosses represent the epoch with the highest accuracy, at which point the model's weights were saved for the final model. Each column is representative case of the types of history curve: no training conducted (left), poor performance and/or lack of improvement on the validation set (middle), and effective training (right).
• Figure 3: Accuracy, F1-score, and AUC distributions of 1000 model instances of Fiducial3 and Fiducial28. Pink and green dashed lines represent the median of each distribution.
• Figure 4: Comparison of model performance across different hyperparameter configurations as well as data augmentations. For the brief description for each model, please refer to Table \ref{['tab:accall']}. The black dot represents the median, and the error bars show the 16--84 percentile range.