Teaching Responsibilities 2017/18:
Research Interests: Monte Carlo Methods, Stochastic Gradient Methods, Stochastic Processes
@Students: Projects in Machine Learning and Computational Statistics are available
- The True Cost of Stochastic Gradient Langevin Dynamics
Tigran Nagapetyan, Andrew B. Duncan, Leonard Hasenclever, Sebastian J. Vollmer, Lukasz Szpruch, Konstantinos Zygalakis
- Stochastic gradient are a key gradient for many machine learning algorithms. This publication shows that there use for rigorous Bayesian inference is questionable?
- Measuring Sample Quality with Diffusions
Jack Gorham, Andrew B. Duncan, Sebastian J. Vollmer, Lester Mackey
- Contribution in statistics and machine learning: Rigorous a posteriori error bounds for a concrete sample from Monte Carlo computations.
- Contribution in mathematics: New regularity results for Poisson equations.
- The Data Study Group brings together researchers and industry to work on data science challenges posed by top companies. Up to six different organisations will each present a different real world data challenge of their choice. You will choose which challenge is of interest to you, working collaboratively in multidisciplinary teams to address the task at hand.
- You can register your contact details if you are interested in being contacted for future Data Study Groups.
- I have been involved in selecting in supervising data science startup's in the Winton accelerator.
- EPSRC First Grant EP/N000188/1 (£100k )
- LRF Grant on Piecewise Markov Processes (£120k )
- LRF Grant for Data Study Group on Data Centric Engineering (£50k), see blog post
- EDRF Grant to support SME engagment in data study group (£150k )
Teaching Responsibilities 2017/18: None
Personal Homepage:vollmer.ms/sebastian (outdated)