{
  "id": "2301.09633",
  "version": "v1",
  "published": "2023-01-23T18:59:28.000Z",
  "updated": "2023-01-23T18:59:28.000Z",
  "title": "Prediction-Powered Inference",
  "authors": [
    "Anastasios N. Angelopoulos",
    "Stephen Bates",
    "Clara Fannjiang",
    "Michael I. Jordan",
    "Tijana Zrnic"
  ],
  "comment": "Code is available at https://github.com/aangelopoulos/prediction-powered-inference",
  "categories": [
    "stat.ML",
    "cs.AI",
    "cs.LG",
    "q-bio.QM",
    "stat.ME"
  ],
  "abstract": "We introduce prediction-powered inference $\\unicode{x2013}$ a framework for performing valid statistical inference when an experimental data set is supplemented with predictions from a machine-learning system such as AlphaFold. Our framework yields provably valid conclusions without making any assumptions on the machine-learning algorithm that supplies the predictions. Higher accuracy of the predictions translates to smaller confidence intervals, permitting more powerful inference. Prediction-powered inference yields simple algorithms for computing valid confidence intervals for statistical objects such as means, quantiles, and linear and logistic regression coefficients. We demonstrate the benefits of prediction-powered inference with data sets from proteomics, genomics, electronic voting, remote sensing, census analysis, and ecology.",
  "revisions": [
    {
      "version": "v1",
      "updated": "2023-01-23T18:59:28.000Z"
    }
  ],
  "analyses": {
    "keywords": [
      "data set",
      "framework yields provably valid conclusions",
      "prediction-powered inference yields simple algorithms",
      "logistic regression coefficients",
      "computing valid confidence intervals"
    ],
    "tags": [
      "github project"
    ],
    "note": {
      "typesetting": "TeX",
      "pages": 0,
      "language": "en",
      "license": "arXiv",
      "status": "editable"
    }
  }
}