Skip to main content Skip to navigation

previous meetings of the Bayesian Optimisation Reading Group


Hoai Phuong Le presented a paper about an algorithm for Bayesian Optimisation problems with input uncertainty. The algorithm proposes building the Gaussian Process model over probability distributions.


The group discussed with professor Chun-Hung Chen (George Mason University) about every member's current work.


The group discussed the Facebook NeverGrad competition 2020 and established a framework for their own work and submission. A private Github has been created for those wishing to contribute.


Paul Kent presented his recent work on Bayesian Optimisation for two player competitive zero-sum games over continuous gamespace. His current work focused on establishing if Equilibrium existed and how to find them.


Michael Pearce presented github and overleaf for effective collaboration.


Hoai Phuong Le presented 2 approachs for computing Knowledge Gradient: discretising the input space X and simulation-based method (discretising over variable Z \sim \mathcal{N}(0,1)) with some comparison experiements.


Jean-Baptiste Mouret (Inria), author of 'Bayesian Optimization with Automatic Prior Selection for Data-Efficient Direct Policy Search' met with the group and discussed his recent work and possible collaborations.


Juan Ungredda presented MCMC techniques for Bayesian Optimisation. How does Maximum Likelihood estimation breakdown and how can 'going Bayesian' help.


Michael Pearce presented some papers from the field of Operational Research that could benefit from Bayesian Optimisation and gave his insights on the best way to improve on the approaches contained within.


Paul Kent Introduced CoCo to the group, a general framework for comparing optimisation algorithms.