Deep Evidential Regression
This addresses the need for efficient and scalable uncertainty learning in safety-critical domains, representing a novel method for a known bottleneck.
The paper tackles the problem of uncertainty estimation in deterministic neural networks for safety-critical applications by proposing a method that learns both aleatoric and epistemic uncertainty without sampling or out-of-distribution training, achieving well-calibrated uncertainty measures on benchmarks and robustness in tasks like computer vision.
Deterministic neural networks (NNs) are increasingly being deployed in safety critical domains, where calibrated, robust, and efficient measures of uncertainty are crucial. In this paper, we propose a novel method for training non-Bayesian NNs to estimate a continuous target as well as its associated evidence in order to learn both aleatoric and epistemic uncertainty. We accomplish this by placing evidential priors over the original Gaussian likelihood function and training the NN to infer the hyperparameters of the evidential distribution. We additionally impose priors during training such that the model is regularized when its predicted evidence is not aligned with the correct output. Our method does not rely on sampling during inference or on out-of-distribution (OOD) examples for training, thus enabling efficient and scalable uncertainty learning. We demonstrate learning well-calibrated measures of uncertainty on various benchmarks, scaling to complex computer vision tasks, as well as robustness to adversarial and OOD test samples.