📝 SummarXiv

Abstract

Gravitational wave detection requires sophisticated signal processing to identify weak astrophysical signals buried in instrumental noise. Traditional matched filtering approaches face computational challenges with diverse signal morphologies and non-stationary noise. This work presents a deep learning approach combining Continuous Wavelet Transform (CWT) preprocessing with Long Short-Term Memory (LSTM) autoencoder architecture for gravitational wave detection in synthetic data. The CWT provides optimal time-frequency decomposition capturing chirp evolution and transient characteristics essential for compact binary coalescence identification. The LSTM autoencoder learns compressed representations while maintaining sensitivity to subtle signal features distinguishing true astrophysical events from noise artifacts. We generate realistic synthetic datasets incorporating binary black hole merger signals with masses ranging from 10 to 80 solar masses, embedded in colored Gaussian noise representative of Advanced LIGO sensitivity. The trained model demonstrates strong performance metrics: 92.3 percent precision, 67.6 percent recall, and 80.6 percent AUC-ROC, with an average precision score of 0.780. These results exceed LIGO's stringent detection thresholds for confident gravitational wave identification. Compared to traditional approaches, the CWT-LSTM autoencoder shows superior ability to maintain low false alarm rates while preserving sensitivity to weak signals. The method's end-to-end learning eliminates hand-crafted features and template banks, offering a promising pathway toward more robust gravitational wave detection systems. The unsupervised nature enables discovery of signals with unknown morphologies, providing complementary "blind search" capability for detecting exotic astrophysical sources and novel physics beyond current theoretical models.