On the Performance of Direct Loss Minimization for Bayesian Neural Networks
This work addresses the performance of pseudo-Bayesian methods for practitioners in Bayesian deep learning, showing incremental insights.
The paper investigates Direct Loss Minimization (DLM) for Bayesian neural networks, finding that it does not improve over ELBO optimization, contrary to prior results for other models.
Direct Loss Minimization (DLM) has been proposed as a pseudo-Bayesian method motivated as regularized loss minimization. Compared to variational inference, it replaces the loss term in the evidence lower bound (ELBO) with the predictive log loss, which is the same loss function used in evaluation. A number of theoretical and empirical results in prior work suggest that DLM can significantly improve over ELBO optimization for some models. However, as we point out in this paper, this is not the case for Bayesian neural networks (BNNs). The paper explores the practical performance of DLM for BNN, the reasons for its failure and its relationship to optimizing the ELBO, uncovering some interesting facts about both algorithms.