Machine Learning for HI Fraction Prediction
A project using Bayesian Physics-Informed Neural Networks to predict the CNM, UNM, and WNM fractions of HI spectra. Specifically looking at a hybrid CNN-Transformer architecture with Bayesian CNN weights and biases for stochastic predictions to provide error bounds on predicted values. Currently refining the architecture and investigating skip connections, as well as the need for Bayesian Transformer weights.
Feb 1, 2024
Dust Attenuation in SAMI & MAGPI
A project looking at the correlations between dust attenuation and galaxy properties (stellar mass, star formation rate, and gas-phase metallicity) in SAMI and MAGPI galaxies, on both spatially-resolved and integrated scales.
Jan 30, 2023