Probabilistic Conformal Prediction Using Conditional Random Samples

Zhendong Wang, Ruijiang Gao, Mingzhang Yin, Mingyuan Zhou, David Blei
Proceedings of The 26th International Conference on Artificial Intelligence and Statistics, PMLR 206:8814-8836, 2023.

Abstract

This paper proposes probabilistic conformal prediction (PCP), a predictive inference algorithm that estimates a target variable by a discontinuous predictive set. Given inputs, PCP constructs the predictive set based on random samples from an estimated generative model. It is efficient and compatible with conditional generative models with either explicit or implicit density functions. We show that PCP guarantees correct marginal coverage with finite samples and give empirical evidence of conditional coverage. We study PCP on a variety of simulated and real datasets. Compared to existing conformal prediction methods, PCP provides sharper predictive sets.

Cite this Paper

BibTeX
@InProceedings{pmlr-v206-wang23n, title = {Probabilistic Conformal Prediction Using Conditional Random Samples}, author = {Wang, Zhendong and Gao, Ruijiang and Yin, Mingzhang and Zhou, Mingyuan and Blei, David}, booktitle = {Proceedings of The 26th International Conference on Artificial Intelligence and Statistics}, pages = {8814--8836}, year = {2023}, editor = {Ruiz, Francisco and Dy, Jennifer and van de Meent, Jan-Willem}, volume = {206}, series = {Proceedings of Machine Learning Research}, month = {25--27 Apr}, publisher = {PMLR}, pdf = {https://proceedings.mlr.press/v206/wang23n/wang23n.pdf}, url = {https://proceedings.mlr.press/v206/wang23n.html}, abstract = {This paper proposes probabilistic conformal prediction (PCP), a predictive inference algorithm that estimates a target variable by a discontinuous predictive set. Given inputs, PCP constructs the predictive set based on random samples from an estimated generative model. It is efficient and compatible with conditional generative models with either explicit or implicit density functions. We show that PCP guarantees correct marginal coverage with finite samples and give empirical evidence of conditional coverage. We study PCP on a variety of simulated and real datasets. Compared to existing conformal prediction methods, PCP provides sharper predictive sets.} }
Endnote
%0 Conference Paper %T Probabilistic Conformal Prediction Using Conditional Random Samples %A Zhendong Wang %A Ruijiang Gao %A Mingzhang Yin %A Mingyuan Zhou %A David Blei %B Proceedings of The 26th International Conference on Artificial Intelligence and Statistics %C Proceedings of Machine Learning Research %D 2023 %E Francisco Ruiz %E Jennifer Dy %E Jan-Willem van de Meent %F pmlr-v206-wang23n %I PMLR %P 8814--8836 %U https://proceedings.mlr.press/v206/wang23n.html %V 206 %X This paper proposes probabilistic conformal prediction (PCP), a predictive inference algorithm that estimates a target variable by a discontinuous predictive set. Given inputs, PCP constructs the predictive set based on random samples from an estimated generative model. It is efficient and compatible with conditional generative models with either explicit or implicit density functions. We show that PCP guarantees correct marginal coverage with finite samples and give empirical evidence of conditional coverage. We study PCP on a variety of simulated and real datasets. Compared to existing conformal prediction methods, PCP provides sharper predictive sets.
APA
Wang, Z., Gao, R., Yin, M., Zhou, M. & Blei, D.. (2023). Probabilistic Conformal Prediction Using Conditional Random Samples. Proceedings of The 26th International Conference on Artificial Intelligence and Statistics, in Proceedings of Machine Learning Research 206:8814-8836 Available from https://proceedings.mlr.press/v206/wang23n.html.

Related Material