A prediction rigidity formalism for low-cost uncertainties in trained neural networks
Regression methods are fundamental for scientific and technological applications. However,
fitted models can be highly unreliable outside of their training domain, and hence the …
fitted models can be highly unreliable outside of their training domain, and hence the …
Approximative Uncertainty in Neural Network Predictions
M Malmström - 2023 - diva-portal.org
Suppose data-driven black-box models, eg, neural networks, should be used as
components in safety-critical systems such as autonomous vehicles. In that case, knowing …
components in safety-critical systems such as autonomous vehicles. In that case, knowing …