{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2019:PQMV4KFBSITFGL5FA3R7FDCDHW","short_pith_number":"pith:PQMV4KFB","schema_version":"1.0","canonical_sha256":"7c195e28a19226532fa506e3f28c433da3d79c1127f31eeffa63774fdcafc108","source":{"kind":"arxiv","id":"1908.07380","version":5},"attestation_state":"computed","paper":{"title":"PAC-Bayes with Backprop","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["stat.ML"],"primary_cat":"cs.LG","authors_text":"Csaba Szepesvari, Omar Rivasplata, Vikram M Tankasali","submitted_at":"2019-08-19T13:27:08Z","abstract_excerpt":"We explore the family of methods \"PAC-Bayes with Backprop\" (PBB) to train probabilistic neural networks by minimizing PAC-Bayes bounds. We present two training objectives, one derived from a previously known PAC-Bayes bound, and a second one derived from a novel PAC-Bayes bound. Both training objectives are evaluated on MNIST and on various UCI data sets. Our experiments show two striking observations: we obtain competitive test set error estimates (~1.4% on MNIST) and at the same time we compute non-vacuous bounds with much tighter values (~2.3% on MNIST) than previous results. These observat"},"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":"1908.07380","kind":"arxiv","version":5},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2019-08-19T13:27:08Z","cross_cats_sorted":["stat.ML"],"title_canon_sha256":"e1698e025b856ccbc13662f69d83579b6a03d8e9f543598677647aacfcd21203","abstract_canon_sha256":"0052229ecdb6587a63fe1fc192d10bd670e3299d7db1a79902a3b3660fc987fb"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T00:09:51.462279Z","signature_b64":"o+O33czjgPTJNc36YbwflLavnNZzgAFsTACAYzcAV3AY4fcCSbvCr2o6OoJi4MeZqnNF6vpAqWhcewQYrpnrAQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"7c195e28a19226532fa506e3f28c433da3d79c1127f31eeffa63774fdcafc108","last_reissued_at":"2026-07-05T00:09:51.461796Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T00:09:51.461796Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"PAC-Bayes with Backprop","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["stat.ML"],"primary_cat":"cs.LG","authors_text":"Csaba Szepesvari, Omar Rivasplata, Vikram M Tankasali","submitted_at":"2019-08-19T13:27:08Z","abstract_excerpt":"We explore the family of methods \"PAC-Bayes with Backprop\" (PBB) to train probabilistic neural networks by minimizing PAC-Bayes bounds. We present two training objectives, one derived from a previously known PAC-Bayes bound, and a second one derived from a novel PAC-Bayes bound. Both training objectives are evaluated on MNIST and on various UCI data sets. Our experiments show two striking observations: we obtain competitive test set error estimates (~1.4% on MNIST) and at the same time we compute non-vacuous bounds with much tighter values (~2.3% on MNIST) than previous results. These observat"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"1908.07380","kind":"arxiv","version":5},"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/1908.07380/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":"1908.07380","created_at":"2026-07-05T00:09:51.461863+00:00"},{"alias_kind":"arxiv_version","alias_value":"1908.07380v5","created_at":"2026-07-05T00:09:51.461863+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.1908.07380","created_at":"2026-07-05T00:09:51.461863+00:00"},{"alias_kind":"pith_short_12","alias_value":"PQMV4KFBSITF","created_at":"2026-07-05T00:09:51.461863+00:00"},{"alias_kind":"pith_short_16","alias_value":"PQMV4KFBSITFGL5F","created_at":"2026-07-05T00:09:51.461863+00:00"},{"alias_kind":"pith_short_8","alias_value":"PQMV4KFB","created_at":"2026-07-05T00:09:51.461863+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":3,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2605.26222","citing_title":"From Privacy to Generalization: Linear Max-Information Bounds for DP-SGD","ref_index":6,"is_internal_anchor":false},{"citing_arxiv_id":"2310.11203","citing_title":"Federated Learning with Nonvacuous Generalisation Bounds","ref_index":49,"is_internal_anchor":false},{"citing_arxiv_id":"2605.15416","citing_title":"Margin-Adaptive Confidence Ranking for Reliable LLM Judgement","ref_index":243,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/PQMV4KFBSITFGL5FA3R7FDCDHW","json":"https://pith.science/pith/PQMV4KFBSITFGL5FA3R7FDCDHW.json","graph_json":"https://pith.science/api/pith-number/PQMV4KFBSITFGL5FA3R7FDCDHW/graph.json","events_json":"https://pith.science/api/pith-number/PQMV4KFBSITFGL5FA3R7FDCDHW/events.json","paper":"https://pith.science/paper/PQMV4KFB"},"agent_actions":{"view_html":"https://pith.science/pith/PQMV4KFBSITFGL5FA3R7FDCDHW","download_json":"https://pith.science/pith/PQMV4KFBSITFGL5FA3R7FDCDHW.json","view_paper":"https://pith.science/paper/PQMV4KFB","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=1908.07380&json=true","fetch_graph":"https://pith.science/api/pith-number/PQMV4KFBSITFGL5FA3R7FDCDHW/graph.json","fetch_events":"https://pith.science/api/pith-number/PQMV4KFBSITFGL5FA3R7FDCDHW/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/PQMV4KFBSITFGL5FA3R7FDCDHW/action/timestamp_anchor","attest_storage":"https://pith.science/pith/PQMV4KFBSITFGL5FA3R7FDCDHW/action/storage_attestation","attest_author":"https://pith.science/pith/PQMV4KFBSITFGL5FA3R7FDCDHW/action/author_attestation","sign_citation":"https://pith.science/pith/PQMV4KFBSITFGL5FA3R7FDCDHW/action/citation_signature","submit_replication":"https://pith.science/pith/PQMV4KFBSITFGL5FA3R7FDCDHW/action/replication_record"}},"created_at":"2026-07-05T00:09:51.461863+00:00","updated_at":"2026-07-05T00:09:51.461863+00:00"}