{
  "id": "1807.00906",
  "version": "v1",
  "published": "2018-07-02T21:49:32.000Z",
  "updated": "2018-07-02T21:49:32.000Z",
  "title": "Uncertainty in the Variational Information Bottleneck",
  "authors": [
    "Alexander A. Alemi",
    "Ian Fischer",
    "Joshua V. Dillon"
  ],
  "comment": "10 pages, 7 figures. Accepted to UAI 2018 - Uncertainty in Deep Learning Workshop",
  "categories": [
    "cs.LG",
    "stat.ML"
  ],
  "abstract": "We present a simple case study, demonstrating that Variational Information Bottleneck (VIB) can improve a network's classification calibration as well as its ability to detect out-of-distribution data. Without explicitly being designed to do so, VIB gives two natural metrics for handling and quantifying uncertainty.",
  "revisions": [
    {
      "version": "v1",
      "updated": "2018-07-02T21:49:32.000Z"
    }
  ],
  "analyses": {
    "keywords": [
      "variational information bottleneck",
      "uncertainty",
      "simple case study",
      "detect out-of-distribution data",
      "networks classification calibration"
    ],
    "note": {
      "typesetting": "TeX",
      "pages": 10,
      "language": "en",
      "license": "arXiv",
      "status": "editable"
    }
  }
}