Alice Desmons

PhD Student

University of New South Wales

Biography

My main interest is galaxy evolution, particularly the use of tidal features to detect merging galaxies. I'm also interested in machine learning, particularly self-supervised learning. My PhD project is at the intersection of these two interests. Tidal features are faint, diffuse, non-uniform regions of stars that extend into space from a galaxy which indicate that a merger is occuring or has occured within the last few billion years. These features can be used to study samples of merging galaxies and thus give us valuable insight into the galaxy evolution process. To conduct such a study we require a large enough sample of galaxies with tidal features, such that we can analyse properties of merging galaxies over a complete sample and draw accurate conclusions about galaxy evolution. Assembling such a sample is a challenging task for three reasons: tidal feature incidence, the low surface brightness of tidal features, and the lack of automatic classification mechanisms. With big imaging surveys reaching new limiting depths, such as the Vera C Rubin observatory’s Large Survey of Space and Time (LSST) due to commence soon, assembling a large sample of galaxies with tidal features is staring to appear more feasible. However the amount of data predicted to be output from such surveys is virtually impossible to classify visually by humans, even by using large community based projects such as Galaxy Zoo, and hence we are in urgent need of a tool that can automate this classification task and isolate galaxies with tidal features. In this PhD project, I aim to construct a tool to automate detection and classification of tidal features using a method that requires minimal labelled data, known as self-supervised machine learning.