Cedric Ewen

By combining fundamental research in machine learning with applications to the natural sciences, new discoveries can be significantly accelerated. Gravitational-wave science, in particular, relies on fast and accurate parameter estimation methods to extract astrophysical information from observed signals. Recent developments leveraging simulation-based inference have opened new possibilities for machine learning in this field, yet the rapidly growing data volume and the computational cost of high-fidelity simulations, particularly waveform models, create a pressing need for faster, more accurate, and scalable simulation and inference methods. This work addresses these challenges through the development, adaptation, and application of state-of-the-art machine learning methods, including probabilistic AI, generative modeling, simulation-based inference, and physics-informed learning, with the goal of accelerating gravitational-wave discovery while providing broadly applicable tools for complex scientific inference.