1

Bayesian Neural Network Priors Revisited

We show that empirical weight distributions of SGD-trained neural networks are heavy-tailed and correlated and that incorporating these insights into Bayesian neural network priors can improve their performance and reduce the cold-posterior effect.

Vincent Fortuin, Adrià Garriga-Alonso, Florian Wenzel, Gunnar Rätsch, Richard E Turner, Mark van der Wilk, Laurence Aitchison

Data augmentation in Bayesian neural networks and the cold posterior effect

Seth Nabarro, Stoil Ganev, Adrià Garriga-Alonso, Vincent Fortuin, Mark van der Wilk, Laurence Aitchison

Invariance Learning in Deep Neural Networks with Differentiable Laplace Approximations

Alexander Immer, Tycho van der Ouderaa, Vincent Fortuin, Gunnar Rätsch, Mark van der Wilk

On Interpretable Reranking-Based Dependency Parsing Systems

Florian Schottmann, Vincent Fortuin, Edoardo Ponti, Ryan Cotterell

Probing as Quantifying Inductive Bias

Alexander Immer, Lucas Torroba Hennigen, Vincent Fortuin, Ryan Cotterell

PACOH: Bayes-Optimal Meta-Learning with PAC-Guarantees

We derive a novel PAC-Bayes bound for meta-learning with Bayesian models, which gives rise to a computationally efficient meta-learning method that outperforms existing approaches on a range of tasks, especially when the number of meta-tasks is small.

Jonas Rothfuss, Vincent Fortuin, Martin Josifoski, Andreas Krause

Repulsive Deep Ensembles are Bayesian

We show that introducing a repulsive force between the members of a deep ensemble can improve the ensemble’s diversity and performance, especially when this force is applied in the function space, and that it can also guarantee asymptotic convergence to the true Bayes posterior.

Francesco D'Angelo, Vincent Fortuin

Repulsive Deep Ensembles are Bayesian

Scalable Gaussian Process Variational Autoencoders

Metod Jazbec, Matthew Ashman, Vincent Fortuin, Michael Pearce, Stephan Mandt, Gunnar Rätsch

Scalable Marginal Likelihood Estimation for Model Selection in Deep Learning

We show that a Laplace-Generalized-Gauss-Newton approximation to the marginal likelihood of Bayesian neural networks can effectively be used for model selection and can often discover better hyperparameter settings than cross-validation.

Alexander Immer, Matthias Bauer, Vincent Fortuin, Gunnar Rätsch, Mohammad Emtiyaz Khan

T-DPSOM: An Interpretable Clustering Method for Unsupervised Learning of Patient Health States

Laura Manduchi, Matthias Hüser, Martin Faltys, Julia Vogt, Gunnar Rätsch, Vincent Fortuin