Exploring the Frontier of Transient Gravitational Wave Detection: Unleashing the Power of Machine Learning

Abstract

In 2015, the LIGO-Virgo collaboration confirmed the first direct observation of gravitational waves resulting from the merger of two black holes. However, the journey to detect these waves, predicted by Albert Einstein nearly a century earlier, has been arduous. It required decades of technological innovation, persistent scientific investigation, and global collaboration to finally enable us to probe the densest and most energetic regions of the universe. Over the past three observing runs, the second-generation detectors, Advanced LIGO and Advanced Virgo, have confirmed over 90 gravitational wave signals, with many more expected during the ongoing fourth observing run. As the sensitivity of these second-generation detectors continues to improve and third-generation interferometers are added to the network, the number of detections is expected to increase exponentially. Nevertheless, as the number of transient gravitational wave detections grows, new computational challenges will emerge, and novel frontiers will need exploration. Nowadays, machine learning is a booming enterprise. In the past few years, they have gained significant interest due to their success in various tasks and domains, such as their ability to uncover intricate patterns and perform rapid, accurate inference. In this regard, the field of gravitational waves is no exception. In the context of an exploding field of both gravitational wave discoveries and machine learning applications, in this thesis we unleash the power of machine learning to complement state-of-the-art gravitational wave searches and even enhance their performance, fostering a synergistic relationship between the two disciplines.