Statistical machine learning research

Song Liu is looking for good PhD candidates to work together on statistical machine learning research. In general, the student is expected to utilise statistical/mathematical tools to design and understand machine learning algorithms. Song's current research interests are: 

Learning Interpretable Changes from High-dimensional Time-series Data 

The change-points in time-series (such as stock prices) indicate potential phase-transition points of the underlying system and has an important value to data science. However, there has been increasing demands of analyzing time-series with very high dimensionality (such as Neuroimaging data or Gene Expression data) recently. Classic statistical modelling approaches fail terribly on these datasets. This project explores methodologies of analyzing and interpreting changes on these high-dimensional datasets using latest ideas from high-dimensional statistics [1]. 

[1] Bühlmann, P., van de Geer, S., Statistics for High-Dimensional Data: Methods, Theory and Applications, Springer Series in Statistics, 2011. 

Beat Machine Learning by Machine Learning 

It is important to understand machine learning algorithm from an adversarial point of view [1,2]. The robustness of machine learning has become an increasing concern as some systems operate on an adversarial environment: Your opponents are very happy to see your ML system breaking down and they will try to make it happen. To build a robust system, we need to understand how these potential attacks may happen and investigate potential defence mechanisms. 

[1] Mei S. and Zhu X., Using Machine Teaching to Identify Optimal Training-Set Attacks on Machine Learners. In The Twenty-Ninth AAAI Conference on Artificial Intelligence (AAAI), 2015. http://pages.cs.wisc.edu/~jerryzhu/pub/Mei2015Machine.pdf 

[2] Goodfellow, I. J., Shlens, J., & Szegedy, C. (2014). Explaining and harnessing adversarial examples. arXiv preprint arXiv:1412.6572. https://arxiv.org/pdf/1412.6572.pdf 

Song Liu also welcome students who wants to work on other domains of machine learning, such as transfer learning, probabilistic graphical models and topological data analysis. Potential candidates are encouraged to contact Song Liu via his email (song.liu@bristol.ac.uk) before application. 

Eligibility 

The applicant of a PhD under Song Liu’s supervision is expected to have a solid background on multivariate calculus, linear algebra and probability/statistics. 

Applicants should have (or expect to obtain by the start date) at least a good 2.1 degree or equivalent from an overseas institution. 

Funding Opportunities 

Song Liu is currently a fellow of Alan Turing Institute, and can supervise PhD students under Alan Turing Institute PhD studentship. Department of Computer Science and University of Bristol also offer various funding opportunities for domestic and international students. 

Bio of Song Liu 

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.