{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2023:PYFM33S2OZQB4ARA36PJVUQQ35","short_pith_number":"pith:PYFM33S2","schema_version":"1.0","canonical_sha256":"7e0acdee5a76601e0220df9e9ad210df74a12c9e7426f507d7a1be4f6312d9a9","source":{"kind":"arxiv","id":"2304.01119","version":2},"attestation_state":"computed","paper":{"title":"Improved Convergence in High Probability of Clipped Gradient Methods with Heavy Tails","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"math.OC","authors_text":"Alina Ene, Huy L. Nguyen, Ta Duy Nguyen","submitted_at":"2023-04-03T16:34:11Z","abstract_excerpt":"In this work, we study the convergence \\emph{in high probability} of clipped gradient methods when the noise distribution has heavy tails, ie., with bounded $p$th moments, for some $1<p\\le2$. Prior works in this setting follow the same recipe of using concentration inequalities and an inductive argument with union bound to bound the iterates across all iterations. This method results in an increase in the failure probability by a factor of $T$, where $T$ is the number of iterations. We instead propose a new analysis approach based on bounding the moment generating function of a well chosen sup"},"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":"2304.01119","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"math.OC","submitted_at":"2023-04-03T16:34:11Z","cross_cats_sorted":[],"title_canon_sha256":"b05a721c4ebef0c4343e9a387a4a92ad5dc9f7a1cef98525bf40f99655cbdaf8","abstract_canon_sha256":"c3a3e399af4cfb4fa16854bf167fe2169a48a2865110dfd3ee915cda48c9bab0"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T05:58:11.763587Z","signature_b64":"EOiNkMI/RQHfcKGgA2dUP6NK/n6Eb+8iMzQnq3+ORxrSHN0oUuopI2xkcYLyWQUDSdsejwewuBxt7o8Fv9iHBQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"7e0acdee5a76601e0220df9e9ad210df74a12c9e7426f507d7a1be4f6312d9a9","last_reissued_at":"2026-07-05T05:58:11.763222Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T05:58:11.763222Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Improved Convergence in High Probability of Clipped Gradient Methods with Heavy Tails","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"math.OC","authors_text":"Alina Ene, Huy L. Nguyen, Ta Duy Nguyen","submitted_at":"2023-04-03T16:34:11Z","abstract_excerpt":"In this work, we study the convergence \\emph{in high probability} of clipped gradient methods when the noise distribution has heavy tails, ie., with bounded $p$th moments, for some $1<p\\le2$. Prior works in this setting follow the same recipe of using concentration inequalities and an inductive argument with union bound to bound the iterates across all iterations. This method results in an increase in the failure probability by a factor of $T$, where $T$ is the number of iterations. We instead propose a new analysis approach based on bounding the moment generating function of a well chosen sup"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2304.01119","kind":"arxiv","version":2},"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/2304.01119/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":"2304.01119","created_at":"2026-07-05T05:58:11.763275+00:00"},{"alias_kind":"arxiv_version","alias_value":"2304.01119v2","created_at":"2026-07-05T05:58:11.763275+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2304.01119","created_at":"2026-07-05T05:58:11.763275+00:00"},{"alias_kind":"pith_short_12","alias_value":"PYFM33S2OZQB","created_at":"2026-07-05T05:58:11.763275+00:00"},{"alias_kind":"pith_short_16","alias_value":"PYFM33S2OZQB4ARA","created_at":"2026-07-05T05:58:11.763275+00:00"},{"alias_kind":"pith_short_8","alias_value":"PYFM33S2","created_at":"2026-07-05T05:58:11.763275+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2605.15388","citing_title":"Unified High-Probability Analysis of Stochastic Variance-Reduced Estimation","ref_index":109,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/PYFM33S2OZQB4ARA36PJVUQQ35","json":"https://pith.science/pith/PYFM33S2OZQB4ARA36PJVUQQ35.json","graph_json":"https://pith.science/api/pith-number/PYFM33S2OZQB4ARA36PJVUQQ35/graph.json","events_json":"https://pith.science/api/pith-number/PYFM33S2OZQB4ARA36PJVUQQ35/events.json","paper":"https://pith.science/paper/PYFM33S2"},"agent_actions":{"view_html":"https://pith.science/pith/PYFM33S2OZQB4ARA36PJVUQQ35","download_json":"https://pith.science/pith/PYFM33S2OZQB4ARA36PJVUQQ35.json","view_paper":"https://pith.science/paper/PYFM33S2","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2304.01119&json=true","fetch_graph":"https://pith.science/api/pith-number/PYFM33S2OZQB4ARA36PJVUQQ35/graph.json","fetch_events":"https://pith.science/api/pith-number/PYFM33S2OZQB4ARA36PJVUQQ35/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/PYFM33S2OZQB4ARA36PJVUQQ35/action/timestamp_anchor","attest_storage":"https://pith.science/pith/PYFM33S2OZQB4ARA36PJVUQQ35/action/storage_attestation","attest_author":"https://pith.science/pith/PYFM33S2OZQB4ARA36PJVUQQ35/action/author_attestation","sign_citation":"https://pith.science/pith/PYFM33S2OZQB4ARA36PJVUQQ35/action/citation_signature","submit_replication":"https://pith.science/pith/PYFM33S2OZQB4ARA36PJVUQQ35/action/replication_record"}},"created_at":"2026-07-05T05:58:11.763275+00:00","updated_at":"2026-07-05T05:58:11.763275+00:00"}