Multivariate Deep Evidential Regression

We discuss three issues with a proposed solution to extract aleatoric and epistemic model uncertainty from regression-based neural networks (NN). The aforementioned proposal derives a technique by placing evidential priors over the original Gaussian likelihood function and training the NN to infer the hyperparemters of the evidential distribution. Doing so allows for the simultaneous extraction of both uncertainties without sampling or utilization of out-of-distribution data for univariate regression tasks. We describe our issues in detail, give a possible solution and generalize the technique for the multivariate case.