{
  "id": "2207.08200",
  "version": "v1",
  "published": "2022-07-17T15:10:25.000Z",
  "updated": "2022-07-17T15:10:25.000Z",
  "title": "Uncertainty Calibration in Bayesian Neural Networks via Distance-Aware Priors",
  "authors": [
    "Gianluca Detommaso",
    "Alberto Gasparin",
    "Andrew Wilson",
    "Cedric Archambeau"
  ],
  "categories": [
    "stat.ML",
    "cs.AI",
    "cs.LG"
  ],
  "abstract": "As we move away from the data, the predictive uncertainty should increase, since a great variety of explanations are consistent with the little available information. We introduce Distance-Aware Prior (DAP) calibration, a method to correct overconfidence of Bayesian deep learning models outside of the training domain. We define DAPs as prior distributions over the model parameters that depend on the inputs through a measure of their distance from the training set. DAP calibration is agnostic to the posterior inference method, and it can be performed as a post-processing step. We demonstrate its effectiveness against several baselines in a variety of classification and regression problems, including benchmarks designed to test the quality of predictive distributions away from the data.",
  "revisions": [
    {
      "version": "v1",
      "updated": "2022-07-17T15:10:25.000Z"
    }
  ],
  "analyses": {
    "keywords": [
      "bayesian neural networks",
      "distance-aware prior",
      "uncertainty calibration",
      "bayesian deep learning models outside",
      "posterior inference method"
    ],
    "note": {
      "typesetting": "TeX",
      "pages": 0,
      "language": "en",
      "license": "arXiv",
      "status": "editable"
    }
  }
}