{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2021:HEARIDW6UMCLFNIWKCXKIU5FII","short_pith_number":"pith:HEARIDW6","schema_version":"1.0","canonical_sha256":"3901140edea304b2b51650aea453a542237cb3a6336cf450e9efa931705b442d","source":{"kind":"arxiv","id":"2106.04149","version":6},"attestation_state":"computed","paper":{"title":"To Smooth or Not? When Label Smoothing Meets Noisy Labels","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Gang Niu, Hangyu Liu, Jiaheng Wei, Masashi Sugiyama, Tongliang Liu, Yang Liu","submitted_at":"2021-06-08T07:32:29Z","abstract_excerpt":"Label smoothing (LS) is an arising learning paradigm that uses the positively weighted average of both the hard training labels and uniformly distributed soft labels. It was shown that LS serves as a regularizer for training data with hard labels and therefore improves the generalization of the model. Later it was reported LS even helps with improving robustness when learning with noisy labels. However, we observed that the advantage of LS vanishes when we operate in a high label noise regime. Intuitively speaking, this is due to the increased entropy of $\\mathbb{P}(\\text{noisy label}|X)$ when"},"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":"2106.04149","kind":"arxiv","version":6},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2021-06-08T07:32:29Z","cross_cats_sorted":[],"title_canon_sha256":"1047a0e87f60b960787e33ad9eec16764be87142bdf4a5210d17ec392dca6568","abstract_canon_sha256":"1bfc2b98e7c66242bb3b4881900656a5044a85f41d67ef7d5a97d02e124e4ce6"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T04:34:45.733746Z","signature_b64":"uU7aQKlV/p/m/w7NkvYHYrLCKLE5JKOuAOUjRXBxD4Qtxc7kZR6Vh0Yr8VKL+cyvDg3rXK0OF0bpKOgz9WN3DQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"3901140edea304b2b51650aea453a542237cb3a6336cf450e9efa931705b442d","last_reissued_at":"2026-07-05T04:34:45.733244Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T04:34:45.733244Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"To Smooth or Not? When Label Smoothing Meets Noisy Labels","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Gang Niu, Hangyu Liu, Jiaheng Wei, Masashi Sugiyama, Tongliang Liu, Yang Liu","submitted_at":"2021-06-08T07:32:29Z","abstract_excerpt":"Label smoothing (LS) is an arising learning paradigm that uses the positively weighted average of both the hard training labels and uniformly distributed soft labels. It was shown that LS serves as a regularizer for training data with hard labels and therefore improves the generalization of the model. Later it was reported LS even helps with improving robustness when learning with noisy labels. However, we observed that the advantage of LS vanishes when we operate in a high label noise regime. Intuitively speaking, this is due to the increased entropy of $\\mathbb{P}(\\text{noisy label}|X)$ when"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2106.04149","kind":"arxiv","version":6},"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/2106.04149/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":"2106.04149","created_at":"2026-07-05T04:34:45.733303+00:00"},{"alias_kind":"arxiv_version","alias_value":"2106.04149v6","created_at":"2026-07-05T04:34:45.733303+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2106.04149","created_at":"2026-07-05T04:34:45.733303+00:00"},{"alias_kind":"pith_short_12","alias_value":"HEARIDW6UMCL","created_at":"2026-07-05T04:34:45.733303+00:00"},{"alias_kind":"pith_short_16","alias_value":"HEARIDW6UMCLFNIW","created_at":"2026-07-05T04:34:45.733303+00:00"},{"alias_kind":"pith_short_8","alias_value":"HEARIDW6","created_at":"2026-07-05T04:34:45.733303+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.19998","citing_title":"Tri-Info: Generalizable, Interpretable Failure Prediction for VLA Models via Information Theory","ref_index":246,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/HEARIDW6UMCLFNIWKCXKIU5FII","json":"https://pith.science/pith/HEARIDW6UMCLFNIWKCXKIU5FII.json","graph_json":"https://pith.science/api/pith-number/HEARIDW6UMCLFNIWKCXKIU5FII/graph.json","events_json":"https://pith.science/api/pith-number/HEARIDW6UMCLFNIWKCXKIU5FII/events.json","paper":"https://pith.science/paper/HEARIDW6"},"agent_actions":{"view_html":"https://pith.science/pith/HEARIDW6UMCLFNIWKCXKIU5FII","download_json":"https://pith.science/pith/HEARIDW6UMCLFNIWKCXKIU5FII.json","view_paper":"https://pith.science/paper/HEARIDW6","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2106.04149&json=true","fetch_graph":"https://pith.science/api/pith-number/HEARIDW6UMCLFNIWKCXKIU5FII/graph.json","fetch_events":"https://pith.science/api/pith-number/HEARIDW6UMCLFNIWKCXKIU5FII/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/HEARIDW6UMCLFNIWKCXKIU5FII/action/timestamp_anchor","attest_storage":"https://pith.science/pith/HEARIDW6UMCLFNIWKCXKIU5FII/action/storage_attestation","attest_author":"https://pith.science/pith/HEARIDW6UMCLFNIWKCXKIU5FII/action/author_attestation","sign_citation":"https://pith.science/pith/HEARIDW6UMCLFNIWKCXKIU5FII/action/citation_signature","submit_replication":"https://pith.science/pith/HEARIDW6UMCLFNIWKCXKIU5FII/action/replication_record"}},"created_at":"2026-07-05T04:34:45.733303+00:00","updated_at":"2026-07-05T04:34:45.733303+00:00"}