{"paper":{"title":"PPI++: Efficient Prediction-Powered Inference","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"PPI++ yields confidence sets for any parameter dimension that always improve on classical intervals by adapting to the quality of machine learning predictions on unlabeled data.","cross_cats":["cs.LG","stat.ME"],"primary_cat":"stat.ML","authors_text":"Anastasios N. Angelopoulos, John C. Duchi, Tijana Zrnic","submitted_at":"2023-11-02T17:59:04Z","abstract_excerpt":"We present PPI++: a computationally lightweight methodology for estimation and inference based on a small labeled dataset and a typically much larger dataset of machine-learning predictions. The methods automatically adapt to the quality of available predictions, yielding easy-to-compute confidence sets -- for parameters of any dimensionality -- that always improve on classical intervals using only the labeled data. PPI++ builds on prediction-powered inference (PPI), which targets the same problem setting, improving its computational and statistical efficiency. Real and synthetic experiments d"},"claims":{"count":4,"items":[{"kind":"strongest_claim","text":"PPI++ builds on prediction-powered inference (PPI), which targets the same problem setting, improving its computational and statistical efficiency. Real and synthetic experiments demonstrate the benefits of the proposed adaptations.","source":"verdict.strongest_claim","status":"machine_extracted","claim_id":"C1","attestation":"unclaimed"},{"kind":"weakest_assumption","text":"The method can automatically adapt to the quality of available predictions in a way that guarantees improvement over classical intervals for parameters of any dimensionality.","source":"verdict.weakest_assumption","status":"machine_extracted","claim_id":"C2","attestation":"unclaimed"},{"kind":"one_line_summary","text":"PPI++ yields easy-to-compute confidence sets for any-dimensional parameters that always improve on classical intervals from labeled data alone by leveraging abundant ML predictions.","source":"verdict.one_line_summary","status":"machine_extracted","claim_id":"C3","attestation":"unclaimed"},{"kind":"headline","text":"PPI++ yields confidence sets for any parameter dimension that always improve on classical intervals by adapting to the quality of machine learning predictions on unlabeled data.","source":"verdict.pith_extraction.headline","status":"machine_extracted","claim_id":"C4","attestation":"unclaimed"}],"snapshot_sha256":"c3b25330d487e036300c066af64369a45906d757c6a0f76a8e74b62d05839c3d"},"source":{"id":"2311.01453","kind":"arxiv","version":2},"verdict":{"id":"76122b95-082e-4919-9901-d429037992a8","model_set":{"reader":"grok-4.3"},"created_at":"2026-05-17T12:17:59.886611Z","strongest_claim":"PPI++ builds on prediction-powered inference (PPI), which targets the same problem setting, improving its computational and statistical efficiency. Real and synthetic experiments demonstrate the benefits of the proposed adaptations.","one_line_summary":"PPI++ yields easy-to-compute confidence sets for any-dimensional parameters that always improve on classical intervals from labeled data alone by leveraging abundant ML predictions.","pipeline_version":"pith-pipeline@v0.9.0","weakest_assumption":"The method can automatically adapt to the quality of available predictions in a way that guarantees improvement over classical intervals for parameters of any dimensionality.","pith_extraction_headline":"PPI++ yields confidence sets for any parameter dimension that always improve on classical intervals by adapting to the quality of machine learning predictions on unlabeled data."},"references":{"count":25,"sample":[{"doi":"","year":2023,"title":"Angelopoulos, Stephen Bates, Clara Fannjiang, Michael I","work_id":"c8193bfd-7617-4fa3-a20f-cfcdf275a345","ref_index":1,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2023,"title":"A. N. Angelopoulos, J. C. Duchi, and T. Zrnic. A note on statistical efficiency in Prediction-Powered Inference. 2023. URL https://web.stanford.edu/~jduchi/projects/ AngelopoulosDuZr23w.pdf","work_id":"a603ce3f-91ba-4e14-b445-a2128667281a","ref_index":2,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":1998,"title":"P. Bickel, C. A. J. Klaassen, Y. Ritov, and J. Wellner. Efficient and Adaptive Estimation for Semiparametric Models. Springer Verlag, 1998","work_id":"3aceaaac-5ed0-404f-b394-bed549dd4a49","ref_index":3,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2004,"title":"S. Boyd and L. Vandenberghe. Convex Optimization. Cambridge University Press, 2004. 18","work_id":"be7bb34f-546f-4bb8-98b1-f6143f39e45f","ref_index":4,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":1986,"title":"L. D. Brown. Fundamentals of Statistical Exponential Families . Institute of Mathematical Statistics, Hayward, California, 1986","work_id":"2ae6e995-0e38-4867-8c86-4fb731ff4bf1","ref_index":5,"cited_arxiv_id":"","is_internal_anchor":false}],"resolved_work":25,"snapshot_sha256":"6c68683fe23fec48b61c7f2f0899b777041349b6aed6da7a60b82909277c12d2","internal_anchors":0},"formal_canon":{"evidence_count":2,"snapshot_sha256":"f9b73baa44ef20ee2945954de28ad7ab907c287d02328a8ee61b2690dd4673f1"},"author_claims":{"count":0,"strong_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"builder_version":"pith-number-builder-2026-05-17-v1"}