{
  "abstract": "Diffusion models can sometimes reproduce examples from their training data, a phenomenon known as memorization. Existing memorization detection methods for unconditional diffusion models typically operate at a pre-selected timestep and therefore cannot adapt the detection time to the evolving denoising trajectory. Sequential monitoring enables adaptive detection but repeated testing across timesteps complicates trajectory-level false-alarm control. We study two ways of combining split conformal calibration with sequential monitoring for prompt-free memorization detection: per-step Bonferroni calibration and direct trajectory-level calibration of the maximum memorization score. The former controls repeated testing by splitting the false-alarm budget across timesteps, whereas the latter calibrates the event of ever crossing a threshold directly through the trajectory-level maximum. We evaluate this formulation on an unconditional CIFAR-10 EDM using two complementary per-step statistics based on geometry and representation spikiness. Experiments show that sequential monitoring improves the power-timing tradeoff over fixed-time detection. Compared with Bonferroni calibration, trajectory-level calibration achieves comparable detection power while using the false-alarm budget less conservatively and enabling earlier detection at higher noise levels.",
  "abstract_shown": true,
  "affiliations": [],
  "article_type": null,
  "arxiv": "2610.34617",
  "author_total": 4,
  "authors": [
    {
      "name": "Yixiao Jing"
    },
    {
      "name": "Chia-Yu Hsu"
    },
    {
      "name": "Milad Hoseinpour"
    },
    {
      "name": "Zekai Zhang"
    }
  ],
  "bibtex": "https://research.draconitic.com/paper/3ep1dv2g5tvnb.bib",
  "cited_by": 0,
  "doi": "10.48550/arxiv.2610.34617",
  "fulltext_url": null,
  "id": "3ep1dv2g5tvnb",
  "issns": [],
  "landing_url": "https://arxiv.org/abs/2610.34617",
  "language": null,
  "licence": "CC-BY-4.0",
  "licence_label": "CC BY 4.0",
  "licence_source": "article",
  "licence_url": "http://creativecommons.org/licenses/by/4.0/",
  "metadata_licence": "CC0-1.0",
  "native_subjects": [
    [
      "arxiv",
      "cs.LG"
    ],
    [
      "arxiv",
      "stat.ML"
    ],
    [
      "openalex",
      "1702|Adversarial Robustness in Machine Learning"
    ]
  ],
  "openalex": "W7223862875",
  "pdf_url": "https://arxiv.org/pdf/2610.34617",
  "pmcid": null,
  "pmid": null,
  "published": "2026-10-25",
  "publisher": null,
  "ris": "https://research.draconitic.com/paper/3ep1dv2g5tvnb.ris",
  "sources": [
    {
      "first_seen": "2026-10-30T12:35:49Z",
      "native_id": "2610.34617",
      "source": "arxiv",
      "url": "https://arxiv.org/abs/2610.34617"
    }
  ],
  "subjects": [
    {
      "field": "mathematics",
      "origin": "arxiv:stat.ML",
      "subfield": "statistics"
    }
  ],
  "title": "Valid Early Detection of Memorization in Diffusion Models",
  "title_text": "Valid Early Detection of Memorization in Diffusion Models",
  "type": "preprint",
  "updated_at": "2026-10-30T12:35:49Z",
  "url": "https://research.draconitic.com/paper/3ep1dv2g5tvnb",
  "venue": "arXiv",
  "versions": []
}