{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2019:MB4ZSWCT2I5G4IBFJWFCBIFROR","short_pith_number":"pith:MB4ZSWCT","schema_version":"1.0","canonical_sha256":"6079995853d23a6e20254d8a20a0b174784bf14237d3f7629b3a1aa1b2f560e9","source":{"kind":"arxiv","id":"1906.02629","version":3},"attestation_state":"computed","paper":{"title":"When Does Label Smoothing Help?","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["stat.ML"],"primary_cat":"cs.LG","authors_text":"Geoffrey Hinton, Rafael M\\\"uller, Simon Kornblith","submitted_at":"2019-06-06T15:03:11Z","abstract_excerpt":"The generalization and learning speed of a multi-class neural network can often be significantly improved by using soft targets that are a weighted average of the hard targets and the uniform distribution over labels. Smoothing the labels in this way prevents the network from becoming over-confident and label smoothing has been used in many state-of-the-art models, including image classification, language translation and speech recognition. Despite its widespread use, label smoothing is still poorly understood. Here we show empirically that in addition to improving generalization, label smooth"},"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":"1906.02629","kind":"arxiv","version":3},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2019-06-06T15:03:11Z","cross_cats_sorted":["stat.ML"],"title_canon_sha256":"bb52780f59dfc53ac3a28454c7f246ef2b1e3f811dfa95902177a841ebeaa963","abstract_canon_sha256":"a9b03b5739218f066e591b5705f5f5085494c6b885915f480ead7603d5744c11"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T01:09:19.240400Z","signature_b64":"1IVSrM/LkCCBa/0To7wBG4UygJuqDz5Hf5T7m5Wafori+L9v1CywqBs914AzwNKQTPqb9MzmM5pZ64GJwEUfDg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"6079995853d23a6e20254d8a20a0b174784bf14237d3f7629b3a1aa1b2f560e9","last_reissued_at":"2026-07-05T01:09:19.239890Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T01:09:19.239890Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"When Does Label Smoothing Help?","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["stat.ML"],"primary_cat":"cs.LG","authors_text":"Geoffrey Hinton, Rafael M\\\"uller, Simon Kornblith","submitted_at":"2019-06-06T15:03:11Z","abstract_excerpt":"The generalization and learning speed of a multi-class neural network can often be significantly improved by using soft targets that are a weighted average of the hard targets and the uniform distribution over labels. Smoothing the labels in this way prevents the network from becoming over-confident and label smoothing has been used in many state-of-the-art models, including image classification, language translation and speech recognition. Despite its widespread use, label smoothing is still poorly understood. Here we show empirically that in addition to improving generalization, label smooth"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"1906.02629","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/1906.02629/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":"1906.02629","created_at":"2026-07-05T01:09:19.239947+00:00"},{"alias_kind":"arxiv_version","alias_value":"1906.02629v3","created_at":"2026-07-05T01:09:19.239947+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.1906.02629","created_at":"2026-07-05T01:09:19.239947+00:00"},{"alias_kind":"pith_short_12","alias_value":"MB4ZSWCT2I5G","created_at":"2026-07-05T01:09:19.239947+00:00"},{"alias_kind":"pith_short_16","alias_value":"MB4ZSWCT2I5G4IBF","created_at":"2026-07-05T01:09:19.239947+00:00"},{"alias_kind":"pith_short_8","alias_value":"MB4ZSWCT","created_at":"2026-07-05T01:09:19.239947+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":6,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.02507","citing_title":"Towards Automated Discovery: A Review of Generative Models, Multimodal Learning and Closed-Loop Workflows in Inverse Materials Design","ref_index":83,"is_internal_anchor":false},{"citing_arxiv_id":"2606.09881","citing_title":"Toward Calibrated, Fair, and accurate Deepfake Detection","ref_index":50,"is_internal_anchor":false},{"citing_arxiv_id":"2605.23249","citing_title":"Enhancing Deep Neural Network Reliability with Refinement and Calibration","ref_index":10,"is_internal_anchor":false},{"citing_arxiv_id":"2605.15416","citing_title":"Margin-Adaptive Confidence Ranking for Reliable LLM Judgement","ref_index":37,"is_internal_anchor":false},{"citing_arxiv_id":"2509.20786","citing_title":"LiLAW: Lightweight Learnable Adaptive Weighting to Learn Sample Difficulty & Improve Noisy Training","ref_index":17,"is_internal_anchor":false},{"citing_arxiv_id":"2605.12753","citing_title":"Optimization in Sparse 2D to Dense 3D Weakly Supervised Learning: Application to Multi-Label Segmentation of Large ex vivo MRI Data","ref_index":8,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/MB4ZSWCT2I5G4IBFJWFCBIFROR","json":"https://pith.science/pith/MB4ZSWCT2I5G4IBFJWFCBIFROR.json","graph_json":"https://pith.science/api/pith-number/MB4ZSWCT2I5G4IBFJWFCBIFROR/graph.json","events_json":"https://pith.science/api/pith-number/MB4ZSWCT2I5G4IBFJWFCBIFROR/events.json","paper":"https://pith.science/paper/MB4ZSWCT"},"agent_actions":{"view_html":"https://pith.science/pith/MB4ZSWCT2I5G4IBFJWFCBIFROR","download_json":"https://pith.science/pith/MB4ZSWCT2I5G4IBFJWFCBIFROR.json","view_paper":"https://pith.science/paper/MB4ZSWCT","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=1906.02629&json=true","fetch_graph":"https://pith.science/api/pith-number/MB4ZSWCT2I5G4IBFJWFCBIFROR/graph.json","fetch_events":"https://pith.science/api/pith-number/MB4ZSWCT2I5G4IBFJWFCBIFROR/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/MB4ZSWCT2I5G4IBFJWFCBIFROR/action/timestamp_anchor","attest_storage":"https://pith.science/pith/MB4ZSWCT2I5G4IBFJWFCBIFROR/action/storage_attestation","attest_author":"https://pith.science/pith/MB4ZSWCT2I5G4IBFJWFCBIFROR/action/author_attestation","sign_citation":"https://pith.science/pith/MB4ZSWCT2I5G4IBFJWFCBIFROR/action/citation_signature","submit_replication":"https://pith.science/pith/MB4ZSWCT2I5G4IBFJWFCBIFROR/action/replication_record"}},"created_at":"2026-07-05T01:09:19.239947+00:00","updated_at":"2026-07-05T01:09:19.239947+00:00"}