{
  "id": "2405.15006",
  "version": "v1",
  "published": "2024-05-23T19:23:09.000Z",
  "updated": "2024-05-23T19:23:09.000Z",
  "title": "Path-metrics, pruning, and generalization",
  "authors": [
    "Antoine Gonon",
    "Nicolas Brisebarre",
    "Elisa Riccietti",
    "Rémi Gribonval"
  ],
  "categories": [
    "cs.LG",
    "stat.ML"
  ],
  "abstract": "Analyzing the behavior of ReLU neural networks often hinges on understanding the relationships between their parameters and the functions they implement. This paper proves a new bound on function distances in terms of the so-called path-metrics of the parameters. Since this bound is intrinsically invariant with respect to the rescaling symmetries of the networks, it sharpens previously known bounds. It is also, to the best of our knowledge, the first bound of its kind that is broadly applicable to modern networks such as ResNets, VGGs, U-nets, and many more. In contexts such as network pruning and quantization, the proposed path-metrics can be efficiently computed using only two forward passes. Besides its intrinsic theoretical interest, the bound yields not only novel theoretical generalization bounds, but also a promising proof of concept for rescaling-invariant pruning.",
  "revisions": [
    {
      "version": "v1",
      "updated": "2024-05-23T19:23:09.000Z"
    }
  ],
  "analyses": {
    "keywords": [
      "path-metrics",
      "relu neural networks",
      "novel theoretical generalization bounds",
      "function distances",
      "parameters"
    ],
    "note": {
      "typesetting": "TeX",
      "pages": 0,
      "language": "en",
      "license": "arXiv",
      "status": "editable"
    }
  }
}