{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2020:ZLXKEUV4ZGYEKSFS3WQIYS4D4P","short_pith_number":"pith:ZLXKEUV4","schema_version":"1.0","canonical_sha256":"caeea252bcc9b04548b2dda08c4b83e3ceb9b7f3d09b5a96c391b8f1d6d3b544","source":{"kind":"arxiv","id":"2011.09161","version":3},"attestation_state":"computed","paper":{"title":"Positive-Congruent Training: Towards Regression-Free Model Updates","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"cs.CV","authors_text":"Kaustav Kundu, Meng Wang, Shuo Yang, Sijie Yan, Siqi Deng, Stefano Soatto, Wei Xia, Yuanjun Xiong","submitted_at":"2020-11-18T09:00:44Z","abstract_excerpt":"Reducing inconsistencies in the behavior of different versions of an AI system can be as important in practice as reducing its overall error. In image classification, sample-wise inconsistencies appear as \"negative flips\": A new model incorrectly predicts the output for a test sample that was correctly classified by the old (reference) model. Positive-congruent (PC) training aims at reducing error rate while at the same time reducing negative flips, thus maximizing congruency with the reference model only on positive predictions, unlike model distillation. We propose a simple approach for PC t"},"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":"2011.09161","kind":"arxiv","version":3},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CV","submitted_at":"2020-11-18T09:00:44Z","cross_cats_sorted":["cs.LG"],"title_canon_sha256":"4a0298928f53de5bdff635c6845e95f8f8923f4d954e5a504c4f70deb6be7b58","abstract_canon_sha256":"20287553e520a5721cd66ba1485e7f0960306a7787a96158c988fd7c42e35ee8"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T02:40:49.955060Z","signature_b64":"xNaJEOC/qxuFtBcY1dN/+3J+RadTqWvrNbt0x751JhKU0ZS1mHbp6bdrUj0ao/m60dITAos2VWt/RJHUhB10Cg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"caeea252bcc9b04548b2dda08c4b83e3ceb9b7f3d09b5a96c391b8f1d6d3b544","last_reissued_at":"2026-07-05T02:40:49.954491Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T02:40:49.954491Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Positive-Congruent Training: Towards Regression-Free Model Updates","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"cs.CV","authors_text":"Kaustav Kundu, Meng Wang, Shuo Yang, Sijie Yan, Siqi Deng, Stefano Soatto, Wei Xia, Yuanjun Xiong","submitted_at":"2020-11-18T09:00:44Z","abstract_excerpt":"Reducing inconsistencies in the behavior of different versions of an AI system can be as important in practice as reducing its overall error. In image classification, sample-wise inconsistencies appear as \"negative flips\": A new model incorrectly predicts the output for a test sample that was correctly classified by the old (reference) model. Positive-congruent (PC) training aims at reducing error rate while at the same time reducing negative flips, thus maximizing congruency with the reference model only on positive predictions, unlike model distillation. We propose a simple approach for PC t"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2011.09161","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/2011.09161/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":"2011.09161","created_at":"2026-07-05T02:40:49.954567+00:00"},{"alias_kind":"arxiv_version","alias_value":"2011.09161v3","created_at":"2026-07-05T02:40:49.954567+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2011.09161","created_at":"2026-07-05T02:40:49.954567+00:00"},{"alias_kind":"pith_short_12","alias_value":"ZLXKEUV4ZGYE","created_at":"2026-07-05T02:40:49.954567+00:00"},{"alias_kind":"pith_short_16","alias_value":"ZLXKEUV4ZGYEKSFS","created_at":"2026-07-05T02:40:49.954567+00:00"},{"alias_kind":"pith_short_8","alias_value":"ZLXKEUV4","created_at":"2026-07-05T02:40:49.954567+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":1,"sample":[{"citing_arxiv_id":"2411.10896","citing_title":"Targeting Negative Flips in Active Learning using Validation Sets","ref_index":1,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/ZLXKEUV4ZGYEKSFS3WQIYS4D4P","json":"https://pith.science/pith/ZLXKEUV4ZGYEKSFS3WQIYS4D4P.json","graph_json":"https://pith.science/api/pith-number/ZLXKEUV4ZGYEKSFS3WQIYS4D4P/graph.json","events_json":"https://pith.science/api/pith-number/ZLXKEUV4ZGYEKSFS3WQIYS4D4P/events.json","paper":"https://pith.science/paper/ZLXKEUV4"},"agent_actions":{"view_html":"https://pith.science/pith/ZLXKEUV4ZGYEKSFS3WQIYS4D4P","download_json":"https://pith.science/pith/ZLXKEUV4ZGYEKSFS3WQIYS4D4P.json","view_paper":"https://pith.science/paper/ZLXKEUV4","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2011.09161&json=true","fetch_graph":"https://pith.science/api/pith-number/ZLXKEUV4ZGYEKSFS3WQIYS4D4P/graph.json","fetch_events":"https://pith.science/api/pith-number/ZLXKEUV4ZGYEKSFS3WQIYS4D4P/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/ZLXKEUV4ZGYEKSFS3WQIYS4D4P/action/timestamp_anchor","attest_storage":"https://pith.science/pith/ZLXKEUV4ZGYEKSFS3WQIYS4D4P/action/storage_attestation","attest_author":"https://pith.science/pith/ZLXKEUV4ZGYEKSFS3WQIYS4D4P/action/author_attestation","sign_citation":"https://pith.science/pith/ZLXKEUV4ZGYEKSFS3WQIYS4D4P/action/citation_signature","submit_replication":"https://pith.science/pith/ZLXKEUV4ZGYEKSFS3WQIYS4D4P/action/replication_record"}},"created_at":"2026-07-05T02:40:49.954567+00:00","updated_at":"2026-07-05T02:40:49.954567+00:00"}