abstract
- This paper presents a methodology that combines topological data analysis and convolutional neural networks (CNNs) to improve gravitational wave detection. Using persistent homology, topological features are extracted and integrated into a CNN model, improving its ability to distinguish relevant signals in the presence of noise. Experiments show that this approach not only increases signal detection accuracy, but also reduces the computational complexity involved in letting the neural network learn without the aid of persistent homology. These findings open up possibilities for efficient gravitational wave detection in noisy signal analysis. © The Author(s), under exclusive license to Springer Nature Switzerland AG 2026.