{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:YSAKQENDFXTMJW3NGCPMHATY5D","short_pith_number":"pith:YSAKQEND","schema_version":"1.0","canonical_sha256":"c480a811a32de6c4db6d309ec38278e8e136614820618e95a80ff28935a6d114","source":{"kind":"arxiv","id":"2402.16991","version":3},"attestation_state":"computed","paper":{"title":"A Phase Transition in Diffusion Models Reveals the Hierarchical Nature of Data","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cond-mat.dis-nn","cs.CV","cs.LG"],"primary_cat":"stat.ML","authors_text":"Alessandro Favero, Antonio Sclocchi, Matthieu Wyart","submitted_at":"2024-02-26T19:52:33Z","abstract_excerpt":"Understanding the structure of real data is paramount in advancing modern deep-learning methodologies. Natural data such as images are believed to be composed of features organized in a hierarchical and combinatorial manner, which neural networks capture during learning. Recent advancements show that diffusion models can generate high-quality images, hinting at their ability to capture this underlying compositional structure. We study this phenomenon in a hierarchical generative model of data. We find that the backward diffusion process acting after a time $t$ is governed by a phase transition"},"verification_status":{"content_addressed":true,"pith_receipt":true,"author_attested":false,"weak_author_claims":0,"strong_author_claims":0,"externally_anchored":false,"storage_verified":false,"citation_signatures":0,"replication_records":0,"graph_snapshot":true,"references_resolved":false,"formal_links_present":false},"canonical_record":{"source":{"id":"2402.16991","kind":"arxiv","version":3},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"stat.ML","submitted_at":"2024-02-26T19:52:33Z","cross_cats_sorted":["cond-mat.dis-nn","cs.CV","cs.LG"],"title_canon_sha256":"9033fed64b32bfde30f4d40ec7417bfc89a992559ff1c95e87dcb5ce4e4ec705","abstract_canon_sha256":"d02952022dbc0a1ee4728062fdd89bc9bf8f19ad39a06a183aea38a95c8f5439"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T09:53:40.321740Z","signature_b64":"R3B88fVKq8bzVeaVDjGWu/u/NC9Vbz+NeUcxnCUgqyNxc7OhDNNt3RNqmuH4boDYBnrE/SvxUZ5wIqsc3E5hAw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"c480a811a32de6c4db6d309ec38278e8e136614820618e95a80ff28935a6d114","last_reissued_at":"2026-07-05T09:53:40.321294Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T09:53:40.321294Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"A Phase Transition in Diffusion Models Reveals the Hierarchical Nature of Data","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cond-mat.dis-nn","cs.CV","cs.LG"],"primary_cat":"stat.ML","authors_text":"Alessandro Favero, Antonio Sclocchi, Matthieu Wyart","submitted_at":"2024-02-26T19:52:33Z","abstract_excerpt":"Understanding the structure of real data is paramount in advancing modern deep-learning methodologies. Natural data such as images are believed to be composed of features organized in a hierarchical and combinatorial manner, which neural networks capture during learning. Recent advancements show that diffusion models can generate high-quality images, hinting at their ability to capture this underlying compositional structure. We study this phenomenon in a hierarchical generative model of data. We find that the backward diffusion process acting after a time $t$ is governed by a phase transition"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2402.16991","kind":"arxiv","version":3},"verdict":{"id":null,"model_set":{},"created_at":null,"strongest_claim":"","one_line_summary":"","pipeline_version":null,"weakest_assumption":"","pith_extraction_headline":""},"integrity":{"clean":true,"summary":{"advisory":0,"critical":0,"by_detector":{},"informational":0},"endpoint":"/pith/2402.16991/integrity.json","findings":[],"available":true,"detectors_run":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938"},"references":{"count":0,"sample":[],"resolved_work":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57","internal_anchors":0},"formal_canon":{"evidence_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"author_claims":{"count":0,"strong_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"builder_version":"pith-number-builder-2026-05-17-v1"},"aliases":[{"alias_kind":"arxiv","alias_value":"2402.16991","created_at":"2026-07-05T09:53:40.321361+00:00"},{"alias_kind":"arxiv_version","alias_value":"2402.16991v3","created_at":"2026-07-05T09:53:40.321361+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2402.16991","created_at":"2026-07-05T09:53:40.321361+00:00"},{"alias_kind":"pith_short_12","alias_value":"YSAKQENDFXTM","created_at":"2026-07-05T09:53:40.321361+00:00"},{"alias_kind":"pith_short_16","alias_value":"YSAKQENDFXTMJW3N","created_at":"2026-07-05T09:53:40.321361+00:00"},{"alias_kind":"pith_short_8","alias_value":"YSAKQEND","created_at":"2026-07-05T09:53:40.321361+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":5,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.20299","citing_title":"Statistical Properties of Training & Generalization","ref_index":230,"is_internal_anchor":false},{"citing_arxiv_id":"2606.13022","citing_title":"Quality-Preserving Imperceptible Adversarial Attack on Skeleton-based Human Action Recognition","ref_index":75,"is_internal_anchor":false},{"citing_arxiv_id":"2606.20299","citing_title":"Statistical Properties of Training & Generalization","ref_index":230,"is_internal_anchor":false},{"citing_arxiv_id":"2508.06614","citing_title":"Local Diffusion Models and Phases of Data Distributions","ref_index":24,"is_internal_anchor":false},{"citing_arxiv_id":"2604.02415","citing_title":"Generative models on phase space","ref_index":66,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/YSAKQENDFXTMJW3NGCPMHATY5D","json":"https://pith.science/pith/YSAKQENDFXTMJW3NGCPMHATY5D.json","graph_json":"https://pith.science/api/pith-number/YSAKQENDFXTMJW3NGCPMHATY5D/graph.json","events_json":"https://pith.science/api/pith-number/YSAKQENDFXTMJW3NGCPMHATY5D/events.json","paper":"https://pith.science/paper/YSAKQEND"},"agent_actions":{"view_html":"https://pith.science/pith/YSAKQENDFXTMJW3NGCPMHATY5D","download_json":"https://pith.science/pith/YSAKQENDFXTMJW3NGCPMHATY5D.json","view_paper":"https://pith.science/paper/YSAKQEND","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2402.16991&json=true","fetch_graph":"https://pith.science/api/pith-number/YSAKQENDFXTMJW3NGCPMHATY5D/graph.json","fetch_events":"https://pith.science/api/pith-number/YSAKQENDFXTMJW3NGCPMHATY5D/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/YSAKQENDFXTMJW3NGCPMHATY5D/action/timestamp_anchor","attest_storage":"https://pith.science/pith/YSAKQENDFXTMJW3NGCPMHATY5D/action/storage_attestation","attest_author":"https://pith.science/pith/YSAKQENDFXTMJW3NGCPMHATY5D/action/author_attestation","sign_citation":"https://pith.science/pith/YSAKQENDFXTMJW3NGCPMHATY5D/action/citation_signature","submit_replication":"https://pith.science/pith/YSAKQENDFXTMJW3NGCPMHATY5D/action/replication_record"}},"created_at":"2026-07-05T09:53:40.321361+00:00","updated_at":"2026-07-05T09:53:40.321361+00:00"}