{
  "abstract": "Calibrating a conformal prediction set at each early exit does not preserve coverage once a data-dependent exit is selected. We construct one terminal path score from prefix maxima, calibrate a single split-conformal quantile, and return nested prefix sets that all contain the valid terminal set. The guarantee is selection-safe without an alpha/L correction and holds simultaneously for every post-calibration label-free stopping policy that preserves terminal-set containment. We also bound the efficiency cost of the path maximum by the empirical tail of the early-over-terminal score excess. Across eight encoder task-backbone cells, pathwise selection attains macro coverage 0.901 versus 0.814 for naive layerwise selection, with average set size 1.41 and latency fraction 0.71. On CIFAR-100 with a 24-exit ViT-L/16 (n_cal=500), first-eligible stopping at budget b=4 attains coverage 0.902, set size 2.76, and latency fraction 0.57, and the same calibration supports deployment-time budget changes without refitting. In a joint exit-count/calibration-size sweep at n=100, pathwise set size moves only from 3.02 to 3.11 as exits increase from 4 to 24, while Bonferroni degenerates to the full label space for 12 or more exits. Paired comparison against MinSE valid selection at (n,L)=(100,24) gives a set-size difference of 0.80 [0.47, 1.13]. Full-vocabulary decoder experiments remain a negative boundary, with sets of 7,162-9,620 tokens. Pathwise calibration is most useful when candidate exits are many relative to calibration size, when budgets change after calibration, and when shallow heads return actionable sets.",
  "abstract_shown": true,
  "affiliations": [],
  "article_type": null,
  "arxiv": "2610.20773",
  "author_total": 2,
  "authors": [
    {
      "name": "Adria Binte Habib"
    },
    {
      "name": "Ibne Farabi Shihab"
    }
  ],
  "bibtex": "https://research.draconitic.com/paper/fm5f1tb9d78jr.bib",
  "cited_by": 0,
  "doi": "10.48550/arxiv.2610.20773",
  "fulltext_url": null,
  "id": "fm5f1tb9d78jr",
  "issns": [],
  "landing_url": "https://arxiv.org/abs/2610.20773",
  "language": null,
  "licence": "LicenseRef-arXiv-nonexclusive-1.0",
  "licence_label": "arXiv licence",
  "licence_source": "article",
  "licence_url": "http://arxiv.org/licenses/nonexclusive-distrib/1.0/",
  "metadata_licence": "CC0-1.0",
  "native_subjects": [
    [
      "arxiv",
      "cs.LG"
    ],
    [
      "arxiv",
      "stat.ML"
    ],
    [
      "openalex",
      "1707|Advanced Neural Network Applications"
    ]
  ],
  "openalex": "W7222492324",
  "pdf_url": "https://arxiv.org/pdf/2610.20773",
  "pmcid": null,
  "pmid": null,
  "published": "2026-10-14",
  "publisher": null,
  "ris": "https://research.draconitic.com/paper/fm5f1tb9d78jr.ris",
  "sources": [
    {
      "first_seen": "2026-10-30T12:30:00Z",
      "native_id": "2610.20773",
      "source": "arxiv",
      "url": "https://arxiv.org/abs/2610.20773"
    }
  ],
  "subjects": [
    {
      "field": "mathematics",
      "origin": "arxiv:stat.ML",
      "subfield": "statistics"
    }
  ],
  "title": "Selection-Safe Conformal Early Exit: Distribution-Free Coverage under Adaptive Depth",
  "title_text": "Selection-Safe Conformal Early Exit: Distribution-Free Coverage under Adaptive Depth",
  "type": "preprint",
  "updated_at": "2026-10-30T12:30:00Z",
  "url": "https://research.draconitic.com/paper/fm5f1tb9d78jr",
  "venue": "arXiv",
  "versions": []
}