Are Bayesian neural networks intrinsically good at out-of-distribution detection?
This challenges a widely held belief in machine learning, potentially impacting researchers and practitioners relying on BNNs for safety-critical applications like autonomous systems, though it is incremental in questioning existing assumptions rather than proposing a new solution.
The paper tackled the assumption that Bayesian neural networks (BNNs) are inherently good at out-of-distribution (OOD) detection by providing empirical evidence that proper Bayesian inference with common architectures does not necessarily lead to effective OOD detection, showing that uncertainties often fail to reflect the data generating process.
The need to avoid confident predictions on unfamiliar data has sparked interest in out-of-distribution (OOD) detection. It is widely assumed that Bayesian neural networks (BNN) are well suited for this task, as the endowed epistemic uncertainty should lead to disagreement in predictions on outliers. In this paper, we question this assumption and provide empirical evidence that proper Bayesian inference with common neural network architectures does not necessarily lead to good OOD detection. To circumvent the use of approximate inference, we start by studying the infinite-width case, where Bayesian inference can be exact considering the corresponding Gaussian process. Strikingly, the kernels induced under common architectural choices lead to uncertainties that do not reflect the underlying data generating process and are therefore unsuited for OOD detection. Finally, we study finite-width networks using HMC, and observe OOD behavior that is consistent with the infinite-width case. Overall, our study discloses fundamental problems when naively using BNNs for OOD detection and opens interesting avenues for future research.