Variation Due to Regularization Tractably Recovers Bayesian Deep Learning
This addresses the problem of reliable uncertainty estimation for safe decision-making in deep learning applications, representing a novel method for a known bottleneck.
The paper tackles uncertainty quantification in deep neural networks by proposing a method based on variation due to regularization, which recovers the Laplace approximation in the infinitesimal limit and scales tractably with network size while maintaining or improving uncertainty quality.
Uncertainty quantification in deep learning is crucial for safe and reliable decision-making in downstream tasks. Existing methods quantify uncertainty at the last layer or other approximations of the network which may miss some sources of uncertainty in the model. To address this gap, we propose an uncertainty quantification method for large networks based on variation due to regularization. Essentially, predictions that are more (less) sensitive to the regularization of network parameters are less (more, respectively) certain. This principle can be implemented by deterministically tweaking the training loss during the fine-tuning phase and reflects confidence in the output as a function of all layers of the network. We show that regularization variation (RegVar) provides rigorous uncertainty estimates that, in the infinitesimal limit, exactly recover the Laplace approximation in Bayesian deep learning. We demonstrate its success in several deep learning architectures, showing it can scale tractably with the network size while maintaining or improving uncertainty quantification quality. Our experiments across multiple datasets show that RegVar not only identifies uncertain predictions effectively but also provides insights into the stability of learned representations.