{
  "annotation_id": "8e03e3dc-073c-42c2-9841-18f1811dfa3f",
  "date_created": "2026-02-17T05:53:23.734000Z",
  "date_modified": "2026-02-17T05:53:23.734000Z",
  "file_hash": "3c4e93376dbe954a0285450a4375b9fddb7b9d7dd634e72b66ac27f40f9126f6",
  "private": false,
  "record": {
    "abstract": "We establish a lower bound on the eluder dimension of generalised linear model classes, showing that standard eluder dimension-based analysis cannot lead to first-order regret bounds. To address this, we introduce a localisation method for the eluder dimension; our analysis immediately recovers and improves on classic results for Bernoulli bandits, and allows for the first genuine first-order bounds for finite-horizon reinforcement learning tasks with bounded cumulative returns.",
    "arxiv_id": "2601.09825",
    "authors": [
      "Alireza Bakhtiari",
      "Alex Ayoub",
      "Samuel Robertson",
      "David Janz",
      "Csaba Szepesv\u00e1ri"
    ],
    "categories": [
      "cs.LG"
    ],
    "license": "http://creativecommons.org/licenses/by/4.0/",
    "title": "Eluder dimension: localise it!",
    "url": "https://arxiv.org/abs/2601.09825",
    "version": "v1"
  },
  "schema_id": "dorsal/arxiv",
  "source": {
    "execution_id": "70c56a2b-789e-4f1c-ab2b-439919f79782",
    "id": "arXiv Dataset",
    "type": "Model",
    "variant": "snapshot-2026-01-17",
    "version": "0.1.0"
  },
  "user_id": 1000002
}