{
  "id": "1812.05836",
  "version": "v1",
  "published": "2018-12-14T09:28:05.000Z",
  "updated": "2018-12-14T09:28:05.000Z",
  "title": "Rethinking Layer-wise Feature Amounts in Convolutional Neural Network Architectures",
  "authors": [
    "Martin Mundt",
    "Sagnik Majumder",
    "Tobias Weis",
    "Visvanathan Ramesh"
  ],
  "comment": "Accepted at the Critiquing and Correcting Trends in Machine Learning (CRACT) Workshop at the 32nd Conference on Neural Information Processing Systems (NeurIPS 2018)",
  "categories": [
    "cs.LG",
    "cs.CV",
    "stat.ML"
  ],
  "abstract": "We characterize convolutional neural networks with respect to the relative amount of features per layer. Using a skew normal distribution as a parametrized framework, we investigate the common assumption of monotonously increasing feature-counts with higher layers of architecture designs. Our evaluation on models with VGG-type layers on the MNIST, Fashion-MNIST and CIFAR-10 image classification benchmarks provides evidence that motivates rethinking of our common assumption: architectures that favor larger early layers seem to yield better accuracy.",
  "revisions": [
    {
      "version": "v1",
      "updated": "2018-12-14T09:28:05.000Z"
    }
  ],
  "analyses": {
    "keywords": [
      "convolutional neural network architectures",
      "rethinking layer-wise feature amounts",
      "common assumption",
      "skew normal distribution",
      "favor larger early layers"
    ],
    "tags": [
      "conference paper"
    ],
    "note": {
      "typesetting": "TeX",
      "pages": 0,
      "language": "en",
      "license": "arXiv",
      "status": "editable"
    }
  }
}