Conformal prediction (CP) for regression can be challenging, especially when the output
distribution is heteroscedastic, multimodal, or skewed. Some of the issues can be addressed …

Conformal prediction converts nearly any point estimator into a prediction interval under
standard assumptions while ensuring valid coverage. However, the extensive computational …

Sparsity is a highly desired feature in deep neural networks (DNNs) since it ensures
numerical efficiency, improves the interpretability (due to the smaller number of relevant …

This thesis develops new methods and associated coverage guarantees for conformal
inference. We begin by discussing applications of the conformal framework to the online …