"Approximate Conditional Coverage via Neural Model Approximations", Allen Schmaltz and Danielle Rasooly, 2022
ADMIT: A general framework for constructing, constraining, and analyzing point predictions and distribution-free prediction sets for deep neural networks.
https://arxiv.org/abs/2205.14310
Workshop on Distribution-Free Uncertainty Quantification at the Thirty-ninth International Conference on Machine Learning (ICML 2022) (Non-archival). Baltimore, Maryland, July 23, 2022