Selected Recorded Talks

I establish a link between variational and Bayesian sampling schemes and deep Ensembles. In particular, both can be seen as implementations of the Wasserstein Gradient Flow; the former as the a regularised and the latter as an unregularised version.

An overview of the research I have done on generalised Bayesian methods 2019-2021

An overview of my ICML paper in 2020 which formally proves that Continual Learning without forgetting is computationally infeasible.

An overview of Generalised Variational Inference, a family of methods to address robustness in Bayesian machine learning problems