{
  "annotation_id": "56d101ff-4d4a-46ba-a07a-c237a22758d5",
  "date_created": "2026-02-17T05:53:12.263000Z",
  "date_modified": "2026-02-17T05:53:12.263000Z",
  "file_hash": "7d10370d0db71ea8e2bc952364245b57b447e01b1177b20dd32a58a0f032be90",
  "private": false,
  "record": {
    "abstract": "It has been demonstrated in various contexts that monotonicity leads to better explainability in neural networks. However, not every function can be well approximated by a monotone neural network. We demonstrate that monotonicity can still be used in two ways to boost explainability. First, we use an adaptation of the decomposition of a trained ReLU network into two monotone and convex parts, thereby overcoming numerical obstacles from an inherent blowup of the weights in this procedure. Our proposed saliency methods - SplitCAM and SplitLRP - improve on state of the art results on both VGG16 and Resnet18 networks on ImageNet-S across all Quantus saliency metric categories. Second, we exhibit that training a model as the difference between two monotone neural networks results in a system with strong self-explainability properties.",
    "arxiv_id": "2601.07700",
    "authors": [
      "Jakob Paul Zimmermann",
      "Georg Loho"
    ],
    "categories": [
      "cs.CV",
      "cs.LG"
    ],
    "license": "http://arxiv.org/licenses/nonexclusive-distrib/1.0/",
    "title": "Hidden Monotonicity: Explaining Deep Neural Networks via their DC Decomposition",
    "url": "https://arxiv.org/abs/2601.07700",
    "version": "v2"
  },
  "schema_id": "dorsal/arxiv",
  "source": {
    "execution_id": "13b2974d-3256-4d63-9849-5bf4f44be9c5",
    "id": "arXiv Dataset",
    "type": "Model",
    "variant": "snapshot-2026-01-17",
    "version": "0.1.0"
  },
  "user_id": 1000002
}