{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:2BYML5I23MQ7HM62R4HB2FGMFM","short_pith_number":"pith:2BYML5I2","schema_version":"1.0","canonical_sha256":"d070c5f51adb21f3b3da8f0e1d14cc2b30bbc448e8c06233ad06051bdc388dcf","source":{"kind":"arxiv","id":"2507.18372","version":1},"attestation_state":"computed","paper":{"title":"On Reconstructing Training Data From Bayesian Posteriors and Trained Models","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.LG","math.ST","stat.TH"],"primary_cat":"stat.ML","authors_text":"George Wynne","submitted_at":"2025-07-24T12:49:41Z","abstract_excerpt":"Publicly releasing the specification of a model with its trained parameters means an adversary can attempt to reconstruct information about the training data via training data reconstruction attacks, a major vulnerability of modern machine learning methods. This paper makes three primary contributions: establishing a mathematical framework to express the problem, characterising the features of the training data that are vulnerable via a maximum mean discrepancy equivalance and outlining a score matching framework for reconstructing data in both Bayesian and non-Bayesian models, the former is a"},"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":"2507.18372","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"stat.ML","submitted_at":"2025-07-24T12:49:41Z","cross_cats_sorted":["cs.LG","math.ST","stat.TH"],"title_canon_sha256":"5c2695e6098313eb3ea9f8c6b3a26da6214bf6043de10e87bf816ed77fca75e1","abstract_canon_sha256":"16787921dfb61ad984309141e6ede40175895025e4b5bae2bb84f6f41ca7e9ed"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:42:46.894041Z","signature_b64":"qTMFSQSfBfXIld4Nanj6IXp5I8ffMtIi5eBNZbO5xF9hEYBlFosBlJH/uPZYXvKkim7qIYafkh5dxP9L1NQXBw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"d070c5f51adb21f3b3da8f0e1d14cc2b30bbc448e8c06233ad06051bdc388dcf","last_reissued_at":"2026-07-05T11:42:46.893634Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:42:46.893634Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"On Reconstructing Training Data From Bayesian Posteriors and Trained Models","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.LG","math.ST","stat.TH"],"primary_cat":"stat.ML","authors_text":"George Wynne","submitted_at":"2025-07-24T12:49:41Z","abstract_excerpt":"Publicly releasing the specification of a model with its trained parameters means an adversary can attempt to reconstruct information about the training data via training data reconstruction attacks, a major vulnerability of modern machine learning methods. This paper makes three primary contributions: establishing a mathematical framework to express the problem, characterising the features of the training data that are vulnerable via a maximum mean discrepancy equivalance and outlining a score matching framework for reconstructing data in both Bayesian and non-Bayesian models, the former is a"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2507.18372","kind":"arxiv","version":1},"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/2507.18372/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":"2507.18372","created_at":"2026-07-05T11:42:46.893690+00:00"},{"alias_kind":"arxiv_version","alias_value":"2507.18372v1","created_at":"2026-07-05T11:42:46.893690+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2507.18372","created_at":"2026-07-05T11:42:46.893690+00:00"},{"alias_kind":"pith_short_12","alias_value":"2BYML5I23MQ7","created_at":"2026-07-05T11:42:46.893690+00:00"},{"alias_kind":"pith_short_16","alias_value":"2BYML5I23MQ7HM62","created_at":"2026-07-05T11:42:46.893690+00:00"},{"alias_kind":"pith_short_8","alias_value":"2BYML5I2","created_at":"2026-07-05T11:42:46.893690+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":0,"internal_anchor_count":0,"sample":[]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/2BYML5I23MQ7HM62R4HB2FGMFM","json":"https://pith.science/pith/2BYML5I23MQ7HM62R4HB2FGMFM.json","graph_json":"https://pith.science/api/pith-number/2BYML5I23MQ7HM62R4HB2FGMFM/graph.json","events_json":"https://pith.science/api/pith-number/2BYML5I23MQ7HM62R4HB2FGMFM/events.json","paper":"https://pith.science/paper/2BYML5I2"},"agent_actions":{"view_html":"https://pith.science/pith/2BYML5I23MQ7HM62R4HB2FGMFM","download_json":"https://pith.science/pith/2BYML5I23MQ7HM62R4HB2FGMFM.json","view_paper":"https://pith.science/paper/2BYML5I2","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2507.18372&json=true","fetch_graph":"https://pith.science/api/pith-number/2BYML5I23MQ7HM62R4HB2FGMFM/graph.json","fetch_events":"https://pith.science/api/pith-number/2BYML5I23MQ7HM62R4HB2FGMFM/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/2BYML5I23MQ7HM62R4HB2FGMFM/action/timestamp_anchor","attest_storage":"https://pith.science/pith/2BYML5I23MQ7HM62R4HB2FGMFM/action/storage_attestation","attest_author":"https://pith.science/pith/2BYML5I23MQ7HM62R4HB2FGMFM/action/author_attestation","sign_citation":"https://pith.science/pith/2BYML5I23MQ7HM62R4HB2FGMFM/action/citation_signature","submit_replication":"https://pith.science/pith/2BYML5I23MQ7HM62R4HB2FGMFM/action/replication_record"}},"created_at":"2026-07-05T11:42:46.893690+00:00","updated_at":"2026-07-05T11:42:46.893690+00:00"}