Below are some select projects
MSc thesis at Imperial College London
Robust low-rank modelling on tensors: new algorithms and extensive comparisons
Supervisor: Dr Stefanos Zafeiriou
- Thesis in PDF
- ICCV paper arxiv preprint and proceedings One paper co-authored for EUSIPCO 2017.
Awarded the Winton Capital Computing MSc Project Prize (best MSc thesis in Computer Science, 1/188 students)
Robust non-negative dictionary learning
A MATLAB implementation of the algorithm from
Robust Non-Negative Dictionary Learning Qihe Pan, Deguang Kong, Chris Ding and Bin Luo In Proceedings of the 28th conference of the AAAI - 2014
that I wrote for the paper I have submitted to ICCV 2017.
A data science competition
I took part in the Pump it Up competition hosted by DrivenData in May 2016 as part of the recruitment process of a London startup. With about two days of quick hacking at the time, I am currently in the top 7%.
Bayesian optimisation for training RBMs
The purpose of this coursework was to experiment with Restricted Boltzman Machines (RBMs). I took it a step further and used Bayesian optimisation with SMAC to find good hyperparameters, and did a bit of statistical analysis to validate some hypothesis and findings.