{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2022:KBGXTZY37LUGF4EOENZBMZ2CTJ","short_pith_number":"pith:KBGXTZY3","schema_version":"1.0","canonical_sha256":"504d79e71bfae862f08e23721667429a40722d81057253b402b2f26f44f6b7a7","source":{"kind":"arxiv","id":"2202.05189","version":3},"attestation_state":"computed","paper":{"title":"Understanding Rare Spurious Correlations in Neural Networks","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["stat.ML"],"primary_cat":"cs.LG","authors_text":"Chi-Ning Chou, Kamalika Chaudhuri, Yao-Yuan Yang","submitted_at":"2022-02-10T17:44:37Z","abstract_excerpt":"Neural networks are known to use spurious correlations such as background information for classification. While prior work has looked at spurious correlations that are widespread in the training data, in this work, we investigate how sensitive neural networks are to rare spurious correlations, which may be harder to detect and correct, and may lead to privacy leaks. We introduce spurious patterns correlated with a fixed class to a few training examples and find that it takes only a handful of such examples for the network to learn the correlation. Furthermore, these rare spurious correlations "},"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":"2202.05189","kind":"arxiv","version":3},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2022-02-10T17:44:37Z","cross_cats_sorted":["stat.ML"],"title_canon_sha256":"370afeb211b5aa0fa41da9660102915b56359713928f521891170b19241259b6","abstract_canon_sha256":"8bcdd09d86dd320d40f3775c6ab5660f9f4cc449bf24e3baa0e6f143bd0066b1"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T05:03:35.260053Z","signature_b64":"BsdIncbwTlTsG6/8+mfeNKNyM26Qb29Hy6yJpAelhXgivz/gpgjL8E/mJWmZ6qAnnkOCvqQF6A//6AXclMO/CA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"504d79e71bfae862f08e23721667429a40722d81057253b402b2f26f44f6b7a7","last_reissued_at":"2026-07-05T05:03:35.259657Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T05:03:35.259657Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Understanding Rare Spurious Correlations in Neural Networks","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["stat.ML"],"primary_cat":"cs.LG","authors_text":"Chi-Ning Chou, Kamalika Chaudhuri, Yao-Yuan Yang","submitted_at":"2022-02-10T17:44:37Z","abstract_excerpt":"Neural networks are known to use spurious correlations such as background information for classification. While prior work has looked at spurious correlations that are widespread in the training data, in this work, we investigate how sensitive neural networks are to rare spurious correlations, which may be harder to detect and correct, and may lead to privacy leaks. We introduce spurious patterns correlated with a fixed class to a few training examples and find that it takes only a handful of such examples for the network to learn the correlation. Furthermore, these rare spurious correlations "},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2202.05189","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/2202.05189/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":"2202.05189","created_at":"2026-07-05T05:03:35.259714+00:00"},{"alias_kind":"arxiv_version","alias_value":"2202.05189v3","created_at":"2026-07-05T05:03:35.259714+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2202.05189","created_at":"2026-07-05T05:03:35.259714+00:00"},{"alias_kind":"pith_short_12","alias_value":"KBGXTZY37LUG","created_at":"2026-07-05T05:03:35.259714+00:00"},{"alias_kind":"pith_short_16","alias_value":"KBGXTZY37LUGF4EO","created_at":"2026-07-05T05:03:35.259714+00:00"},{"alias_kind":"pith_short_8","alias_value":"KBGXTZY3","created_at":"2026-07-05T05:03:35.259714+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":3,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2605.30462","citing_title":"idSCD: Identifying Training Datasets through Semantic Correlation Descriptors","ref_index":48,"is_internal_anchor":false},{"citing_arxiv_id":"2604.18970","citing_title":"Mechanistic Anomaly Detection via Functional Attribution","ref_index":13,"is_internal_anchor":false},{"citing_arxiv_id":"2605.06865","citing_title":"Dataset Watermarking for Closed LLMs with Provable Detection","ref_index":19,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/KBGXTZY37LUGF4EOENZBMZ2CTJ","json":"https://pith.science/pith/KBGXTZY37LUGF4EOENZBMZ2CTJ.json","graph_json":"https://pith.science/api/pith-number/KBGXTZY37LUGF4EOENZBMZ2CTJ/graph.json","events_json":"https://pith.science/api/pith-number/KBGXTZY37LUGF4EOENZBMZ2CTJ/events.json","paper":"https://pith.science/paper/KBGXTZY3"},"agent_actions":{"view_html":"https://pith.science/pith/KBGXTZY37LUGF4EOENZBMZ2CTJ","download_json":"https://pith.science/pith/KBGXTZY37LUGF4EOENZBMZ2CTJ.json","view_paper":"https://pith.science/paper/KBGXTZY3","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2202.05189&json=true","fetch_graph":"https://pith.science/api/pith-number/KBGXTZY37LUGF4EOENZBMZ2CTJ/graph.json","fetch_events":"https://pith.science/api/pith-number/KBGXTZY37LUGF4EOENZBMZ2CTJ/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/KBGXTZY37LUGF4EOENZBMZ2CTJ/action/timestamp_anchor","attest_storage":"https://pith.science/pith/KBGXTZY37LUGF4EOENZBMZ2CTJ/action/storage_attestation","attest_author":"https://pith.science/pith/KBGXTZY37LUGF4EOENZBMZ2CTJ/action/author_attestation","sign_citation":"https://pith.science/pith/KBGXTZY37LUGF4EOENZBMZ2CTJ/action/citation_signature","submit_replication":"https://pith.science/pith/KBGXTZY37LUGF4EOENZBMZ2CTJ/action/replication_record"}},"created_at":"2026-07-05T05:03:35.259714+00:00","updated_at":"2026-07-05T05:03:35.259714+00:00"}