Accelerated computational discovery and design of novel magnesium alloys by machine learning

This thesis embarked on what was, from conception, an exploration for employing data-driven methodologies to propel the field of magnesium (Mg) alloy design. Unfolding across three distinct dimensions of data science, the research aimed to expedite digital alloy design via creating an open-source Mg alloy database, unveiling new insights, developing predictive models to predict mechanical properties of any arbitrary Mg alloy, introducing active learning techniques with a specific focus on Bayesian optimisation, and exploring generative adversarial networks (GANs). Collectively, these efforts have made some progress in (re) defining the paradigm of Mg alloy design.