Tackling covariate shift with node-based Bayesian neural networks
This work addresses covariate shift for machine learning practitioners using Bayesian neural networks, offering an incremental improvement to node-based BNNs.
The paper tackles covariate shift in Bayesian neural networks by interpreting latent noise variables in node-based BNNs as implicit data perturbations, proposing to increase their entropy during training to improve uncertainty estimation under input corruptions, resulting in enhanced performance on out-of-distribution image classification benchmarks and robustness against noisy labels.
Bayesian neural networks (BNNs) promise improved generalization under covariate shift by providing principled probabilistic representations of epistemic uncertainty. However, weight-based BNNs often struggle with high computational complexity of large-scale architectures and datasets. Node-based BNNs have recently been introduced as scalable alternatives, which induce epistemic uncertainty by multiplying each hidden node with latent random variables, while learning a point-estimate of the weights. In this paper, we interpret these latent noise variables as implicit representations of simple and domain-agnostic data perturbations during training, producing BNNs that perform well under covariate shift due to input corruptions. We observe that the diversity of the implicit corruptions depends on the entropy of the latent variables, and propose a straightforward approach to increase the entropy of these variables during training. We evaluate the method on out-of-distribution image classification benchmarks, and show improved uncertainty estimation of node-based BNNs under covariate shift due to input perturbations. As a side effect, the method also provides robustness against noisy training labels.