Being a Bit Frequentist Improves Bayesian Neural Networks
This addresses a key limitation in Bayesian uncertainty quantification for machine learning practitioners, though it is incremental as it adapts an existing frequentist technique to BNNs.
The paper tackled the underperformance of Bayesian neural networks (BNNs) in uncertainty quantification tasks like out-of-distribution (OOD) detection by hypothesizing that BNNs lack OOD data in training, and showed that incorporating OOD data into BNN training makes them competitive with frequentist methods.
Despite their compelling theoretical properties, Bayesian neural networks (BNNs) tend to perform worse than frequentist methods in classification-based uncertainty quantification (UQ) tasks such as out-of-distribution (OOD) detection. In this paper, based on empirical findings in prior works, we hypothesize that this issue is because even recent Bayesian methods have never considered OOD data in their training processes, even though this "OOD training" technique is an integral part of state-of-the-art frequentist UQ methods. To validate this, we treat OOD data as a first-class citizen in BNN training by exploring four different ways of incorporating OOD data into Bayesian inference. We show in extensive experiments that OOD-trained BNNs are competitive to recent frequentist baselines. This work thus provides strong baselines for future work in Bayesian UQ.