{
  "annotation_id": "f22fcd8d-005e-4613-ab88-f11969df576e",
  "date_created": "2026-03-02T18:01:35.462000Z",
  "date_modified": "2026-03-02T18:01:35.462000Z",
  "file_hash": "d3c2c9f672c0e3f50e55ece12ca2b575cabe9063ea61edd0a3a308931d55b569",
  "private": false,
  "record": {
    "abstract": "We present a method of estimating the gradient of an objective function with\nrespect to the synaptic weights of a spiking neural network. The method works\nby measuring the fluctuations in the objective function in response to dynamic\nperturbation of the membrane conductances of the neurons. It is compatible with\nrecurrent networks of conductance-based model neurons with dynamic synapses.\nThe method can be interpreted as a biologically plausible synaptic learning\nrule, if the dynamic perturbations are generated by a special class of\n``empiric\u0027\u0027 synapses driven by random spike trains from an external source.",
    "arxiv_id": "q-bio/0601028",
    "authors": [
      "Ila R. Fiete",
      "H. Sebastian Seung"
    ],
    "categories": [
      "q-bio.NC"
    ],
    "doi": "10.1103/PhysRevLett.97.048104",
    "journal_ref": "Phys. Rev. Lett. 97, 048104 (2006)",
    "title": "Gradient learning in spiking neural networks by dynamic perturbation of   conductances",
    "url": "https://arxiv.org/abs/q-bio/0601028"
  },
  "schema_id": "dorsal/arxiv",
  "source": {
    "execution_id": "a00cdcc5-49e1-460a-9cc6-803abc110cac",
    "id": "arXiv Dataset IDs",
    "type": "Model",
    "variant": "snapshot-2026-03-01",
    "version": "0.1.0"
  },
  "user_id": 1000002
}