Stein Variational Newton Neural Network Ensembles
This addresses uncertainty quantification in deep learning for practitioners needing faster, more accurate Bayesian methods, though it appears incremental as it modifies existing ensemble approaches.
The paper tackled the problem of deep neural network ensembles inadequately leveraging second-order information for uncertainty quantification by proposing a novel approximate Bayesian inference method that incorporates Stein Variational Newton updates with scalable Hessian approximations. The result demonstrated superior performance with significantly reduced training epochs on diverse regression and classification tasks, enhancing uncertainty quantification and robustness against overfitting.
Deep neural network ensembles are powerful tools for uncertainty quantification, which have recently been re-interpreted from a Bayesian perspective. However, current methods inadequately leverage second-order information of the loss landscape, despite the recent availability of efficient Hessian approximations. We propose a novel approximate Bayesian inference method that modifies deep ensembles to incorporate Stein Variational Newton updates. Our approach uniquely integrates scalable modern Hessian approximations, achieving faster convergence and more accurate posterior distribution approximations. We validate the effectiveness of our method on diverse regression and classification tasks, demonstrating superior performance with a significantly reduced number of training epochs compared to existing ensemble-based methods, while enhancing uncertainty quantification and robustness against overfitting.