An out-of-distribution discriminator based on Bayesian neural network epistemic uncertainty

Neural networks have revolutionized the field of machine learning with increased predictive capability. In addition to improving the predictions of neural networks, there is a simultaneous demand for reliable uncertainty quantification on estimates made by machine learning methods such as neural networks. Bayesian neural networks (BNNs) are an important type of neural network with built-in capability for quantifying uncertainty. This paper discusses aleatoric and epistemic uncertainty in BNNs and how they can be calculated. With an example dataset of images where the goal is to identify the amplitude of an event in the image, it is shown that epistemic uncertainty tends to be lower in images which are well-represented in the training dataset and tends to be high in images which are not well-represented. An algorithm for out-of-distribution (OoD) detection with BNN epistemic uncertainty is introduced along with various experiments demonstrating factors influencing the OoD detection capability in a BNN. The OoD detection capability with epistemic uncertainty is shown to be comparable to the OoD detection in the discriminator network of a generative adversarial network (GAN) with comparable network architecture.

Autocart -- spatially-aware regression trees for ecological and spatial modeling

Many ecological and spatial processes are complex in nature and are not accurately modeled by linear models. Regression trees promise to handle the high-order interactions that are present in ecological and spatial datasets, but fail to produce physically realistic characterizations of the underlying landscape. The "autocart" (autocorrelated regression trees) R package extends the functionality of previously proposed spatial regression tree methods through a spatially aware splitting function and novel adaptive inverse distance weighting method in each terminal node. The efficacy of these autocart models, including an autocart extension of random forest, is demonstrated on multiple datasets. This highlights the ability of autocart to model complex interactions between spatial variables while still providing physically realistic representations of the landscape.