Uncertainty Estimations by Softplus normalization in Bayesian Convolutional Neural Networks with Variational Inference
This work addresses uncertainty quantification for classification tasks in machine learning, offering a method that integrates uncertainty measures and regularization, but it is incremental as it builds on existing Bayesian and variational inference frameworks.
The paper tackles uncertainty estimation in Bayesian convolutional neural networks by introducing a Softplus normalization method to coherently estimate aleatoric and epistemic uncertainty, achieving performance equivalent to frequentist inference on datasets like MNIST, CIFAR-10, and CIFAR-100.
We introduce a novel uncertainty estimation for classification tasks for Bayesian convolutional neural networks with variational inference. By normalizing the output of a Softplus function in the final layer, we estimate aleatoric and epistemic uncertainty in a coherent manner. The intractable posterior probability distributions over weights are inferred by Bayes by Backprop. Firstly, we demonstrate how this reliable variational inference method can serve as a fundamental construct for various network architectures. On multiple datasets in supervised learning settings (MNIST, CIFAR-10, CIFAR-100), this variational inference method achieves performances equivalent to frequentist inference in identical architectures, while the two desiderata, a measure for uncertainty and regularization are incorporated naturally. Secondly, we examine how our proposed measure for aleatoric and epistemic uncertainties is derived and validate it on the aforementioned datasets.