{
  "id": "2102.12959",
  "version": "v1",
  "published": "2021-02-24T12:35:43.000Z",
  "updated": "2021-02-24T12:35:43.000Z",
  "title": "A statistical theory of out-of-distribution detection",
  "authors": [
    "Xi Wang",
    "Laurence Aitchison"
  ],
  "categories": [
    "stat.ML",
    "cs.LG"
  ],
  "abstract": "We introduce a principled approach to detecting out-of-distribution (OOD) data by exploiting a connection to data curation. In data curation, we exclude ambiguous or difficult-to-classify input points from the dataset, and these excluded points are by definition OOD. We can therefore obtain the likelihood for OOD points by using a principled generative model of data-curation initially developed to explain the cold-posterior effect in Bayesian neural networks (Aitchison 2020). This model gives higher OOD probabilities when predictive uncertainty is higher and can be trained using maximum-likelihood jointly over the in-distribution and OOD points. This approach gives superior performance to past methods that did not provide a probability for OOD points, and therefore could not be trained using maximum-likelihood.",
  "revisions": [
    {
      "version": "v1",
      "updated": "2021-02-24T12:35:43.000Z"
    }
  ],
  "analyses": {
    "keywords": [
      "out-of-distribution detection",
      "statistical theory",
      "ood points",
      "data curation",
      "bayesian neural networks"
    ],
    "note": {
      "typesetting": "TeX",
      "pages": 0,
      "language": "en",
      "license": "arXiv",
      "status": "editable"
    }
  }
}