This is a set of three lectures on application of machine-learning (ML) using neural-networks in gravitational-wave physics. The first lecture is a research talk specific to gravitational waves. The content of second and third are general, and can be applied to other domains.
- Research talk on application of ML to gravitational-wave astronomy.
- Basics of neural networks: function approximation property
- Construction and training using pytorch framework: loss functions, backpropagation, training loop.
- Distribution approximation: learning an approximator from samples.
- A brief intro to the pytorch-lightning framework.
- An exercise in likelihood-free inference: training a normalizing flow for posterior estimation of parameters of a line.