{
  "id": "1301.3557",
  "version": "v1",
  "published": "2013-01-16T02:12:07.000Z",
  "updated": "2013-01-16T02:12:07.000Z",
  "title": "Stochastic Pooling for Regularization of Deep Convolutional Neural Networks",
  "authors": [
    "Matthew D. Zeiler",
    "Rob Fergus"
  ],
  "comment": "9 pages",
  "categories": [
    "cs.LG",
    "cs.NE",
    "stat.ML"
  ],
  "abstract": "We introduce a simple and effective method for regularizing large convolutional neural networks. We replace the conventional deterministic pooling operations with a stochastic procedure, randomly picking the activation within each pooling region according to a multinomial distribution, given by the activities within the pooling region. The approach is hyper-parameter free and can be combined with other regularization approaches, such as dropout and data augmentation. We achieve state-of-the-art performance on four image datasets, relative to other approaches that do not utilize data augmentation.",
  "revisions": [
    {
      "version": "v1",
      "updated": "2013-01-16T02:12:07.000Z"
    }
  ],
  "analyses": {
    "keywords": [
      "deep convolutional neural networks",
      "stochastic pooling",
      "regularization",
      "regularizing large convolutional neural networks",
      "data augmentation"
    ],
    "note": {
      "typesetting": "TeX",
      "pages": 9,
      "language": "en",
      "license": "arXiv",
      "status": "editable",
      "adsabs": "2013arXiv1301.3557Z"
    }
  }
}