{
  "id": "1905.12782",
  "version": "v1",
  "published": "2019-05-29T23:34:44.000Z",
  "updated": "2019-05-29T23:34:44.000Z",
  "title": "Active Learning in the Overparameterized and Interpolating Regime",
  "authors": [
    "Mina Karzand",
    "Robert D. Nowak"
  ],
  "comment": "20 pages, 10 figures",
  "categories": [
    "cs.LG",
    "cs.AI",
    "stat.ML"
  ],
  "abstract": "Overparameterized models that interpolate training data often display surprisingly good generalization properties. Specifically, minimum norm solutions have been shown to generalize well in the overparameterized, interpolating regime. This paper introduces a new framework for active learning based on the notion of minimum norm interpolators. We analytically study its properties and behavior in the kernel-based setting and present experimental studies with kernel methods and neural networks. In general, active learning algorithms adaptively select examples for labeling that (1) rule-out as many (incompatible) classifiers as possible at each step and/or (2) discover cluster structure in unlabeled data and label representative examples from each cluster. We show that our new active learning approach based on a minimum norm heuristic automatically exploits both these strategies.",
  "revisions": [
    {
      "version": "v1",
      "updated": "2019-05-29T23:34:44.000Z"
    }
  ],
  "analyses": {
    "keywords": [
      "active learning",
      "interpolating regime",
      "minimum norm heuristic automatically exploits",
      "minimum norm solutions",
      "learning algorithms adaptively select examples"
    ],
    "note": {
      "typesetting": "TeX",
      "pages": 20,
      "language": "en",
      "license": "arXiv",
      "status": "editable"
    }
  }
}