Yichuan Zhang

4th year PhD student in statistical machine learning at ANC,
School of Informatics,
University of Edinburgh.
Email: Y.Zhang-60@sms.ed.ac.uk
Supervisor: Charles Sutton

Research Interests

Extracting knowledge from a world full of uncertainties lies at the heart of machine learning. The uncertainties come from the nature of the world and also our limited capability of receiving and processing data. Bayesian statistics offers an elegant framework to deal with the certainties in many Artificial Intelligence (AI) problems. I'm interested in the machine learning methdologies and applications using Bayesian statistics. From a computational perspective, one of the greatest challenges of applying Bayesian statistics arise from the inference. My PhD is forcused on efficient Bayesian inference methods usign Markov chain Monte Carlo (MCMC) methods for a wide range of machine learning applications.

Publications

Semi-Separable Hamiltonian Monte Carlo for Inference in Bayesian Hierarchical Models. Yichuan Zhang, Charles Sutton. Accepted by Advances in Neural Information Processing Systems (NIPS). 2014 last update at 18/9/2014
Quasi-Newton Markov chain Monte Carlo. Yichuan Zhang, Charles Sutton. In Advances in Neural Information Processing Systems (NIPS). 2011. last update at 23/4/2012
Continuous Relaxations for Discrete Hamiltonian Monte Carlo. Yichuan Zhang, Charles Sutton, Amos Storkey, Zoubin Zoubin Ghahramani. In Advances in Neural Information Processing Systems (NIPS). 2012. (Spotlight 72/1467 = 4.9%) last update at 14/11/2012

Code

BFGS Hamiltonian Monte Carlo last update at 19/4/2012

Some Useful Stuff

Matlab Usage
Radford Neal's blog





Page Hits :