Image Data Augmentation for the TAIGA-IACT Experiment with Conditional Generative Adversarial Networks

TL;DR

The paper tackles imbalanced training data in TAIGA-IACT gamma-ray classification by employing a single conditional GAN (cGAN) conditioned on energy and arrival-direction distance to generate both balanced and imbalanced IACT-like images. A 100-class energy-distance conditioning scheme is implemented, with hexagonal camera images preprocessed into square inputs for the cGAN, achieving 98% gamma-like outputs with high confidence and generation speeds orders of magnitude faster than traditional Monte Carlo simulations. An augmentation algorithm computes per-class counts to approximate a near-uniform energy distribution, enabling rapid generation of balanced datasets. While results demonstrate substantial speedups and promising data quality, some distribution mismatches persist, particularly at energy-range edges, motivating future work on more data, finer class definitions, and alternative imbalance-handling strategies.

Abstract

Modern Imaging Atmospheric Cherenkov Telescopes (IACTs) generate a huge amount of data that must be classified automatically, ideally in real time. Currently, machine learning-based solutions are increasingly being used to solve classification problems. However, these classifiers require proper training data sets to work correctly. The problem with training neural networks on real IACT data is that these data need to be pre-labeled, whereas such labeling is difficult and its results are estimates. In addition, the distribution of incoming events is highly imbalanced. Firstly, there is an imbalance in the types of events, since the number of detected gamma quanta is significantly less than the number of protons. Secondly, the energy distribution of particles of the same type is also imbalanced, since high-energy particles are extremely rare. This imbalance results in poorly trained classifiers that, once trained, do not handle rare events correctly. Using only conventional Monte Carlo event simulation methods to solve this problem is possible, but extremely resource-intensive and time-consuming. To address this issue, we propose to perform data augmentation with artificially generated events of the desired type and energy using conditional generative adversarial networks (cGANs), distinguishing classes by energy values. In the paper, we describe a simple algorithm for generating balanced data sets using cGANs. Thus, the proposed neural network model produces both imbalanced data sets for physical analysis as well as balanced data sets suitable for training other neural networks.

Figures (10)

• Figure 1: Example of a hexagonal image recorded by the TAIGA-IACT detector.
• Figure 2: The energy distribution of the training set.
• Figure 3: Assigning class numbers in a snake-like pattern, $i$ - is an energy class number, $j$ - is a distance class number.
• Figure 4: A hexagonal image and the corresponding square image resulting from preprocessing.
• Figure 5: Architecture of the discriminator.