Parallel sessions

Even though beginner-level machine learning techniques have become available to a wider audience in recent years, these methods often prove insufficient in handling the high dimensionality of the data involved in astrophysical research applications. The jump between a simple application of out-of-the-box techniques and high-level data-driven astrophysical research lies in the ability to design methods that are optimised for the specific problem at hand. This workshop will showcase a sample of advanced methods designed to either solve specific challenging data-analysis problems in astrophysics, or to overcome hindering computational complexity. These techniques will be presented by academics in the context of their research as practical examples of what can be undertaken when simpler approaches fail. This will provide participants that have an intermediate-level understanding of statistical-learning methods and a basic machine-learning implementation experience with tools to write research-grade applications of these methods to astrophysics. In this two-hour workshop, three astrophysical machine-learning problems will be presented in the context of the science question they are addressing from a methods and implementation standpoint. Participants will be able to handle and implement the algorithms presented during the workshop in order to ensure that these techniques can be effectively reused in their own research.

Schedule:

09:00 Hongming Tang “Radio Galaxy Zoo: Giant Radio Galaxy Classification using Multi-Domain Deep Learning”
09:45 Mario Morvan “Modelling astrophysical time series with physics-based deep learning”