{
  "abstract": "Deep Belief Networks (DBNs) learn hierarchical generative models without class supervision. We ask how this unsupervised hierarchy reorganizes data, whether it exposes latent class structure, and whether independently initialized networks discover corresponding internal representations. Across MNIST, Fashion-MNIST, and KMNIST, we follow class geometry and accessibility together with representational dimensionality, input reconstruction, and free generation. We find that global class organization generally strengthens with depth even though labels are never available during learning. Early layers frequently make class identity easier to extract, whereas deeper layers form increasingly compact representations in which related features are learned repeatedly. Global class geometry and classifier accessibility can nevertheless diverge, showing that a more ordered representation is not uniformly more informative at every class boundary. Independently trained DBNs converge to closely related depth-specific representation geometries and neuronal response features, indicating that the learned hierarchy is strongly constrained by the data and learning process. Deeper generative paths also produce broader populations of recognizable free samples while preserving less information about individual inputs. Together, these results show that unsupervised depth follows a reproducible path of representational reorganization in which class structure, accessibility, compression, reconstruction, and generation remain distinct but interacting properties.",
  "abstract_shown": true,
  "affiliations": [
    "Cognitive Computational Neuroscience Group, Pattern Recognition Lab, Friedrich-Alexander-University Erlangen-Nürnberg (FAU), Erlangen, Germany",
    "Neuromodulation and Neuroprosthetics, University Hospital Mannheim, University Heidelberg, Mannheim, Germany",
    "Neuroscience Lab, University Hospital Erlangen, Erlangen, Germany",
    "BG Clinic Ludwigshafen, Ludwigshafen, Germany"
  ],
  "article_type": null,
  "arxiv": null,
  "author_total": 11,
  "authors": [
    {
      "affs": [
        0
      ],
      "name": "Patrick Krauss"
    },
    {
      "affs": [
        1
      ],
      "name": "Patrick Krauss"
    },
    {
      "affs": [
        2
      ],
      "name": "Patrick Krauss"
    },
    {
      "affs": [
        3
      ],
      "name": "Patrick Krauss"
    },
    {
      "affs": [
        0
      ],
      "name": "Achim Schilling"
    },
    {
      "affs": [
        1
      ],
      "name": "Achim Schilling"
    },
    {
      "affs": [
        2
      ],
      "name": "Achim Schilling"
    },
    {
      "affs": [
        3
      ],
      "name": "Achim Schilling"
    },
    {
      "affs": [
        0
      ],
      "name": "Andreas K. Maier"
    },
    {
      "affs": [
        1
      ],
      "name": "Thomas Kinfe"
    },
    {
      "affs": [
        0
      ],
      "name": "Claus Metzner"
    }
  ],
  "bibtex": "https://research.draconitic.com/paper/byfn8mztd9b0q.bib",
  "cited_by": 0,
  "doi": "10.3389/fcpxs.2026.1960494",
  "fulltext_url": "https://www.frontiersin.org/articles/10.3389/fcpxs.2026.1960494/full",
  "id": "byfn8mztd9b0q",
  "issns": [
    "2813-6187"
  ],
  "landing_url": "https://doi.org/10.3389/fcpxs.2026.1960494",
  "language": "en",
  "licence": "CC-BY",
  "licence_label": "CC BY",
  "licence_source": "journal",
  "licence_url": "https://creativecommons.org/licenses/by/4.0/",
  "metadata_licence": null,
  "native_subjects": [
    [
      "lcc",
      "QC1-999"
    ],
    [
      "openalex",
      "1702|Neural Networks and Applications"
    ]
  ],
  "openalex": "W7201866685",
  "pdf_url": "https://www.frontiersin.org/journals/complex-systems/articles/10.3389/fcpxs.2026.1960494/pdf",
  "pmcid": null,
  "pmid": null,
  "published": "2026-10-23",
  "publisher": "Frontiers Media S.A.",
  "ris": "https://research.draconitic.com/paper/byfn8mztd9b0q.ris",
  "sources": [
    {
      "first_seen": "2026-10-30T13:43:57Z",
      "native_id": "9023bcb490c14823a2b8a72248ebb5e4",
      "source": "doaj",
      "url": "https://doi.org/10.3389/fcpxs.2026.1960494"
    }
  ],
  "subjects": [
    {
      "field": "physics",
      "origin": "lcc:QC1-999",
      "subfield": "general"
    }
  ],
  "title": "Convergent evolution in neural representation space: emergent order in deep belief networks",
  "title_text": "Convergent evolution in neural representation space: emergent order in deep belief networks",
  "type": "peer-reviewed",
  "updated_at": "2026-10-30T13:43:57Z",
  "url": "https://research.draconitic.com/paper/byfn8mztd9b0q",
  "venue": "Frontiers in Complex Systems",
  "versions": []
}