Song Liu is a lecturer in University of Bristol. Before, he was a Project Assistant Professor in The Institute of Statistical Mathematics, Tokyo, Japan. he got his Doctor of Engineering degree from Tokyo Institute of Technology supervised by Professor Masashi Sugiyama and was awarded the DC2 Fellowship from Japan Society for the Promotion of Science.
Song Liu focuses on statistical machine learning algorithms identifying and interpreting differences from structured datasets. The datasets collected from real world systems are rarely static and usually evolve over time. Comparing two datasets collected at two different time points or spatial locations may highlight the patterns that change with these physical properties. For example, by comparing twitter keywords patterns before and after a social event, we can see that how the online community responds to a major incident which further helps us understand the change of sentiment toward this event.
His main theoretical research focuses on understanding the statistical properties of difference learning algorithms, such as change-point detection algorithms, and creating statistical methods capturing various patterns that evolve over time/space.
His main application research focuses on applying change detection algorithms on different datasets and reveal specific patterns that may help us understand the underlying system. For example, learning changes between two fMRI images helps us understand how our brain changes its behaviour when given two different tasks.
Achievements and awards
Song Liu was the recipient of the DC2 Fellowship from Japan Society for the Promotion of Science.