Explainability through uncertainty: Trustworthy decision-making with neural networks

arXiv:2403.10168v1 Announce Type: new
Abstract: Uncertainty is a key feature of any machine learning model and is particularly important in neural networks, which tend to be overconfident. This overconfidence is worrying under distribution shifts, where the model performance silently degrades as the data distribution diverges from the training data distribution. Uncertainty estimation offers a solution to overconfident models, communicating when the output should (not) be trusted. Although methods for uncertainty estimation have been developed, they have not been explicitly linked …

abstract arxiv cs.cy cs.lg data decision distribution explainability feature key machine machine learning machine learning model making networks neural networks performance stat.ml through training training data trustworthy type uncertainty