Dr Po-Yen Tung is a skilled and dedicated materials scientist who is committed to advancing the field through modern machine-learning techniques. With his multidisciplinary background, he specialises in solving challenging problems in materials and natural science, including materials discovery and materials characterisation in electron microscopy.
Prior to joining the University of Cambridge, Po-Yen's research focused on materials characterisation and machine learning. During his early PhD studies, he concentrated on understanding the failure phenomenon of rails and wind turbine bearing steels using high-resolution microscopies. From 2020, his research centred on machine learning in materials science. One of Po-Yen's significant contributions is to discovering high-entropy Invar alloys using machine learning, with the aim of serving emerging markets for the transport of liquid hydrogen, ammonia, and natural gas. This study proposed an active learning framework that combines machine learning with density-functional theory, thermodynamic calculations, and experiments, to speed up the design process for high-entropy Invar alloys. Specifically, a generative model (Wasserstein autoencoder) was employed along with a property-informed statistical sampling technique (Markov chain Monte Carlo) for the targeted screening of promising new alloys. This research has significant potential to advance the field of alloy discovery.
Currently, Po-Yen Tung is dedicated to analysing electron microscopic data using machine learning. A notable achievement of Po-Yen Tung is the development of an open-source tool, Spectral Interpretation using Gaussian Mixtures and Autoencoder (SIGMA), which has become a valuable resource for researchers in the field. This tool combines multiple machine-learning techniques such as neural networks, probabilistic clustering, and factor analysis to automate the chemical analysis of electron microscopy datasets in an unsupervised manner.
