arXiv 2101.02703

Distribution-Free, Risk-Controlling Prediction Sets

By Stephen Bates, Anastasios Angelopoulos, et al.

Published 2021-01-07

Wiki summary

Explore the paper's summary, context, and related research on Papiers.

While improving prediction accuracy has been the focus of machine learning in recent years, this alone does not suffice for reliable decision-making. Deploying learning systems in consequential settings also requires calibrating and communicating the uncertainty of predictions. To convey instance-wise uncertainty for prediction tasks, we show how to generate set-valued predictions from a black-box predictor that con…

View the original paper on arXiv