# 'In-Between' Uncertainty in Bayesian Neural Networks

@article{Foong2019InBetweenUI, title={'In-Between' Uncertainty in Bayesian Neural Networks}, author={Andrew Y. K. Foong and Yingzhen Li and Jos{\'e} Miguel Hern{\'a}ndez-Lobato and R. Turner}, journal={ArXiv}, year={2019}, volume={abs/1906.11537} }

We describe a limitation in the expressiveness of the predictive uncertainty estimate given by mean-field variational inference (MFVI), a popular approximate inference method for Bayesian neural networks. In particular, MFVI fails to give calibrated uncertainty estimates in between separated regions of observations. This can lead to catastrophically overconfident predictions when testing on out-of-distribution data. Avoiding such overconfidence is critical for active learning, Bayesian… CONTINUE READING

12 Citations

On the Expressiveness of Approximate Inference in Bayesian Neural Networks

- Mathematics, Computer Science
- 2019

2

Improving predictions of Bayesian neural networks via local linearization

- Computer Science, Mathematics
- 2020

1

Learned Uncertainty-Aware (LUNA) Bases for Bayesian Regression using Multi-Headed Auxiliary Networks

- Mathematics, Computer Science
- 2020

Global inducing point variational posteriors for Bayesian neural networks and deep Gaussian processes

- Mathematics, Computer Science
- 2020

1

Estimating Model Uncertainty of Neural Networks in Sparse Information Form

- Computer Science, Mathematics
- 2020

1

#### References

##### Publications referenced by this paper.

SHOWING 1-10 OF 27 REFERENCES

Probabilistic Backpropagation for Scalable Learning of Bayesian Neural Networks

- Computer Science, Mathematics
- 2015

418

Deep Bayesian Bandits Showdown: An Empirical Comparison of Bayesian Deep Networks for Thompson Sampling

- Mathematics, Computer Science
- 2018

86

Simple and Scalable Predictive Uncertainty Estimation using Deep Ensembles

- Computer Science, Mathematics
- 2017

823

Dropout as a Bayesian Approximation: Representing Model Uncertainty in Deep Learning

- Mathematics, Computer Science
- 2016

2115