{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2020:NZNPDAFXH65YXNXLDFW5HJRTKC","short_pith_number":"pith:NZNPDAFX","schema_version":"1.0","canonical_sha256":"6e5af180b73fbb8bb6eb196dd3a63350a79d84c8f7722befabde1e9315d86de2","source":{"kind":"arxiv","id":"2008.03703","version":1},"attestation_state":"computed","paper":{"title":"What Neural Networks Memorize and Why: Discovering the Long Tail via Influence Estimation","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["stat.ML"],"primary_cat":"cs.LG","authors_text":"Chiyuan Zhang, Vitaly Feldman","submitted_at":"2020-08-09T10:12:28Z","abstract_excerpt":"Deep learning algorithms are well-known to have a propensity for fitting the training data very well and often fit even outliers and mislabeled data points. Such fitting requires memorization of training data labels, a phenomenon that has attracted significant research interest but has not been given a compelling explanation so far. A recent work of Feldman (2019) proposes a theoretical explanation for this phenomenon based on a combination of two insights. First, natural image and data distributions are (informally) known to be long-tailed, that is have a significant fraction of rare and atyp"},"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":"2008.03703","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2020-08-09T10:12:28Z","cross_cats_sorted":["stat.ML"],"title_canon_sha256":"fbde7dc3462f38e67e87785fac85a698fd34c819928b3cefec5790cd68030074","abstract_canon_sha256":"c6e6ce410abd9ffb5d493248e80361e0f963fb1d0e474fea7c2d0cd2063f4f36"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T01:25:44.663144Z","signature_b64":"kB6baTrEVoce9HTJA0GkeabHbloZnOmUIujV7HMWDIXudm7SwWglS9jaPlXIwI9dqU03+BkNRdtvDJxx9cXYDw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"6e5af180b73fbb8bb6eb196dd3a63350a79d84c8f7722befabde1e9315d86de2","last_reissued_at":"2026-07-05T01:25:44.662671Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T01:25:44.662671Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"What Neural Networks Memorize and Why: Discovering the Long Tail via Influence Estimation","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["stat.ML"],"primary_cat":"cs.LG","authors_text":"Chiyuan Zhang, Vitaly Feldman","submitted_at":"2020-08-09T10:12:28Z","abstract_excerpt":"Deep learning algorithms are well-known to have a propensity for fitting the training data very well and often fit even outliers and mislabeled data points. Such fitting requires memorization of training data labels, a phenomenon that has attracted significant research interest but has not been given a compelling explanation so far. A recent work of Feldman (2019) proposes a theoretical explanation for this phenomenon based on a combination of two insights. First, natural image and data distributions are (informally) known to be long-tailed, that is have a significant fraction of rare and atyp"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2008.03703","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/2008.03703/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":"2008.03703","created_at":"2026-07-05T01:25:44.662728+00:00"},{"alias_kind":"arxiv_version","alias_value":"2008.03703v1","created_at":"2026-07-05T01:25:44.662728+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2008.03703","created_at":"2026-07-05T01:25:44.662728+00:00"},{"alias_kind":"pith_short_12","alias_value":"NZNPDAFXH65Y","created_at":"2026-07-05T01:25:44.662728+00:00"},{"alias_kind":"pith_short_16","alias_value":"NZNPDAFXH65YXNXL","created_at":"2026-07-05T01:25:44.662728+00:00"},{"alias_kind":"pith_short_8","alias_value":"NZNPDAFX","created_at":"2026-07-05T01:25:44.662728+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.23591","citing_title":"Quantifying the Agreement Between Data-Influence and Data-Similarity to Understand LLM Behavior","ref_index":36,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/NZNPDAFXH65YXNXLDFW5HJRTKC","json":"https://pith.science/pith/NZNPDAFXH65YXNXLDFW5HJRTKC.json","graph_json":"https://pith.science/api/pith-number/NZNPDAFXH65YXNXLDFW5HJRTKC/graph.json","events_json":"https://pith.science/api/pith-number/NZNPDAFXH65YXNXLDFW5HJRTKC/events.json","paper":"https://pith.science/paper/NZNPDAFX"},"agent_actions":{"view_html":"https://pith.science/pith/NZNPDAFXH65YXNXLDFW5HJRTKC","download_json":"https://pith.science/pith/NZNPDAFXH65YXNXLDFW5HJRTKC.json","view_paper":"https://pith.science/paper/NZNPDAFX","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2008.03703&json=true","fetch_graph":"https://pith.science/api/pith-number/NZNPDAFXH65YXNXLDFW5HJRTKC/graph.json","fetch_events":"https://pith.science/api/pith-number/NZNPDAFXH65YXNXLDFW5HJRTKC/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/NZNPDAFXH65YXNXLDFW5HJRTKC/action/timestamp_anchor","attest_storage":"https://pith.science/pith/NZNPDAFXH65YXNXLDFW5HJRTKC/action/storage_attestation","attest_author":"https://pith.science/pith/NZNPDAFXH65YXNXLDFW5HJRTKC/action/author_attestation","sign_citation":"https://pith.science/pith/NZNPDAFXH65YXNXLDFW5HJRTKC/action/citation_signature","submit_replication":"https://pith.science/pith/NZNPDAFXH65YXNXLDFW5HJRTKC/action/replication_record"}},"created_at":"2026-07-05T01:25:44.662728+00:00","updated_at":"2026-07-05T01:25:44.662728+00:00"}