Extending Prediction-Powered Inference through Conformal Prediction

Daniel Csillag, Pedro Dall’Antonia, Claudio Jose Struchiner, Guilherme Tegoni Goedert
Proceedings of the 43rd International Conference on Machine Learning, PMLR 306:21777-21804, 2026.

Abstract

Prediction-powered inference is a recent methodology for the safe use of black-box ML models to impute missing data, strengthening inference of statistical parameters. However, many applications require strong properties besides valid inference, such as privacy, robustness or validity under continuous distribution shifts; deriving prediction-powered methods with such guarantees is generally an arduous process, and has to be done case by case. In this paper, we resolve this issue by connecting prediction-powered inference with conformal prediction: by performing imputation through a calibrated set-predictor, we attain validity while achieving additional guarantees in a natural manner. We instantiate our procedure for the inference of means, Z- and M-estimation, as well as e-values and e-value-based procedures. Furthermore, in the case of e-values, ours is the first general prediction-powered procedure that operates off-line. We demonstrate these advantages by applying our method on private and time-series data. Both tasks are nontrivial within the standard prediction-powered framework but become natural under our method.

Cite this Paper

BibTeX
@InProceedings{pmlr-v306-csillag26a, title = {Extending Prediction-Powered Inference through Conformal Prediction}, author = {Csillag, Daniel and Dall'Antonia, Pedro and Struchiner, Claudio Jose and Goedert, Guilherme Tegoni}, booktitle = {Proceedings of the 43rd International Conference on Machine Learning}, pages = {21777--21804}, year = {2026}, editor = {Zhang, Tong and Dudik, Miroslav and Jaggi, Martin and Agarwal, Alekh and Li, Sharon and Schuurmans, Dale and Zhu, Jerry and Berkenkamp, Felix and Dong, Hanze and Bietti, Alberto}, volume = {306}, series = {Proceedings of Machine Learning Research}, month = {06--11 Jul}, publisher = {PMLR}, pdf = {https://raw.githubusercontent.com/mlresearch/v306/main/assets/csillag26a/csillag26a.pdf}, url = {https://proceedings.mlr.press/v306/csillag26a.html}, abstract = {Prediction-powered inference is a recent methodology for the safe use of black-box ML models to impute missing data, strengthening inference of statistical parameters. However, many applications require strong properties besides valid inference, such as privacy, robustness or validity under continuous distribution shifts; deriving prediction-powered methods with such guarantees is generally an arduous process, and has to be done case by case. In this paper, we resolve this issue by connecting prediction-powered inference with conformal prediction: by performing imputation through a calibrated set-predictor, we attain validity while achieving additional guarantees in a natural manner. We instantiate our procedure for the inference of means, Z- and M-estimation, as well as e-values and e-value-based procedures. Furthermore, in the case of e-values, ours is the first general prediction-powered procedure that operates off-line. We demonstrate these advantages by applying our method on private and time-series data. Both tasks are nontrivial within the standard prediction-powered framework but become natural under our method.} }
Endnote
%0 Conference Paper %T Extending Prediction-Powered Inference through Conformal Prediction %A Daniel Csillag %A Pedro Dall’Antonia %A Claudio Jose Struchiner %A Guilherme Tegoni Goedert %B Proceedings of the 43rd International Conference on Machine Learning %C Proceedings of Machine Learning Research %D 2026 %E Tong Zhang %E Miroslav Dudik %E Martin Jaggi %E Alekh Agarwal %E Sharon Li %E Dale Schuurmans %E Jerry Zhu %E Felix Berkenkamp %E Hanze Dong %E Alberto Bietti %F pmlr-v306-csillag26a %I PMLR %P 21777--21804 %U https://proceedings.mlr.press/v306/csillag26a.html %V 306 %X Prediction-powered inference is a recent methodology for the safe use of black-box ML models to impute missing data, strengthening inference of statistical parameters. However, many applications require strong properties besides valid inference, such as privacy, robustness or validity under continuous distribution shifts; deriving prediction-powered methods with such guarantees is generally an arduous process, and has to be done case by case. In this paper, we resolve this issue by connecting prediction-powered inference with conformal prediction: by performing imputation through a calibrated set-predictor, we attain validity while achieving additional guarantees in a natural manner. We instantiate our procedure for the inference of means, Z- and M-estimation, as well as e-values and e-value-based procedures. Furthermore, in the case of e-values, ours is the first general prediction-powered procedure that operates off-line. We demonstrate these advantages by applying our method on private and time-series data. Both tasks are nontrivial within the standard prediction-powered framework but become natural under our method.
APA
Csillag, D., Dall’Antonia, P., Struchiner, C.J. & Goedert, G.T.. (2026). Extending Prediction-Powered Inference through Conformal Prediction. Proceedings of the 43rd International Conference on Machine Learning, in Proceedings of Machine Learning Research 306:21777-21804 Available from https://proceedings.mlr.press/v306/csillag26a.html.

Related Material