{
  "id": "1712.05440",
  "version": "v1",
  "published": "2017-12-14T20:31:29.000Z",
  "updated": "2017-12-14T20:31:29.000Z",
  "title": "Nonparametric Neural Networks",
  "authors": [
    "George Philipp",
    "Jaime G. Carbonell"
  ],
  "comment": "ICLR 2017",
  "categories": [
    "cs.LG",
    "cs.GT"
  ],
  "abstract": "Automatically determining the optimal size of a neural network for a given task without prior information currently requires an expensive global search and training many networks from scratch. In this paper, we address the problem of automatically finding a good network size during a single training cycle. We introduce *nonparametric neural networks*, a non-probabilistic framework for conducting optimization over all possible network sizes and prove its soundness when network growth is limited via an L_p penalty. We train networks under this framework by continuously adding new units while eliminating redundant units via an L_2 penalty. We employ a novel optimization algorithm, which we term *adaptive radial-angular gradient descent* or *AdaRad*, and obtain promising results.",
  "revisions": [
    {
      "version": "v1",
      "updated": "2017-12-14T20:31:29.000Z"
    }
  ],
  "analyses": {
    "keywords": [
      "nonparametric neural networks",
      "novel optimization algorithm",
      "prior information",
      "eliminating redundant units",
      "expensive global search"
    ],
    "note": {
      "typesetting": "TeX",
      "pages": 0,
      "language": "en",
      "license": "arXiv",
      "status": "editable"
    }
  }
}