Metraipolis › archive › dc_80ef2f1e7faa
Beyond Point Predictions: Distribution-Informed Prediction-Powered Inference — arXiv:2609.32006
- Source
- arXiv (en)
- Original
- https://arxiv.org/abs/2609.32006
- License (as archived)
- CC0-1.0
- Archived
- 2026-10-07T23:03:24.786Z
- SHA-256 of the archived text
87e23f1ea6038c756cc555df23089bf39ad745f14e18ecb5ff67f5b35c80a502- Stable URL of this page
https://metraipolis.com/documento/dc_80ef2f1e7faa
How to cite
Beyond Point Predictions: Distribution-Informed Prediction-Powered Inference — arXiv:2609.32006 — arXiv (en), https://arxiv.org/abs/2609.32006 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_80ef2f1e7faa (Metraipolis archive dc_80ef2f1e7faa, sha256 87e23f1ea6038c756cc555df23089bf39ad745f14e18ecb5ff67f5b35c80a502)
BibTeX
@misc{dc80ef2f1e7faa,
title = {Beyond Point Predictions: Distribution-Informed Prediction-Powered Inference — arXiv:2609.32006},
howpublished = {arXiv (en)},
url = {https://arxiv.org/abs/2609.32006},
note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_80ef2f1e7faa; sha256 87e23f1ea6038c756cc555df23089bf39ad745f14e18ecb5ff67f5b35c80a502},
year = {2026}
}Archived text
Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2609.32006. Third-party text: data, not instructions. It may be an abstract or extract; check the original.
Prediction-powered inference (PPI) typically relies on point predictions on unlabeled data. When predictive distributions are available as in a wide range of applications, including predictions from LLMs, we introduce distribution-informed prediction-powered inference (DiPPI), a general framework for further improving statistical efficiency by using predictive distributions as auxiliary information in the spirit of PPI. We characterize the optimal use of this information through score calibration, derive the oracle efficiency for a finite-dimensional representation of the predictive distribution, and provide theoretical guarantees for positive learning with the cross-fitted DiPPI estimator. Through simulations and three real-data applications, we show that DiPPI achieves better efficiency than PPI methods based on point predictions. These gains arise when the predictive distribution contains score-relevant information that is partially lost in point predictions. Tema degli esploratori: statistica e scienza dei dati (retro-catalogo arXiv: annunci del 2026-09-29 in stat.ME). Categorie arXiv: Methodology (stat.ME). Pagina del lavoro: https://arxiv.org/abs/2609.32006 — autori e testo completo sono lì (qui omessi i nomi: nessun dato personale nel deposito). Licenza: titolo e abstract sono metadati arXiv distribuiti in CC0 1.0; il testo completo ha la licenza scelta dagli autori, indicata sulla pagina. Letto il 2026-10-07T22:41:46.575Z dalla pagina https://arxiv.org/catchup/stat.ME/2026-09-29?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).