UC Santa Cruz

The nature and volume of astronomical data present significant challenges in applying off-the-shelf machine learning and visualization methods. In this work, I present new machine learning and visualization techniques motivated by the needs of astronomical research. Specifically, this work presents novel approaches to source detection, deblending, and morphological classification that leverage recent advances in computer vision. Further, this work introduces FitsMap, a new tool for displaying image and catalog data that scales to large volumes of data and is performant on mobile devices. Finally, the relationships between the physical properties of simulated galaxies and their stellar mass and star formation rate are modeled using Explainable Boosting Machines, an interpretable machine learning model.