{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:HPZ2TYLUIIQHXHRG4FXIOJQXM6","short_pith_number":"pith:HPZ2TYLU","schema_version":"1.0","canonical_sha256":"3bf3a9e17442207b9e26e16e87261767bc066827e81d9a2eca60b588c355868c","source":{"kind":"arxiv","id":"2605.02701","version":2},"attestation_state":"computed","paper":{"title":"Robust and Fast Training via Per-Sample Clipping","license":"http://creativecommons.org/licenses/by/4.0/","headline":"Per-sample gradient clipping in SGD achieves optimal convergence rates for non-convex problems under heavy-tailed noise.","cross_cats":["cs.LG","stat.ML"],"primary_cat":"math.OC","authors_text":"Davide Nobile, Philipp Grohs","submitted_at":"2026-05-04T15:11:36Z","abstract_excerpt":"We propose a robust gradient estimator based on per-sample gradient clipping and analyze its properties both theoretically and empirically. We show that the resulting method, per-sample clipped SGD (PS-Clip-SGD), achieves optimal in-expectation convergence rates for non-convex optimization problems under heavy-tailed gradient noise. Moreover, we establish high-probability convergence guarantees that match the in-expectation rates up to polylogarithmic factors in the failure probability. We complement our theoretical results with multiple numerical experiments. In particular, we demonstrate tha"},"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":true},"canonical_record":{"source":{"id":"2605.02701","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"math.OC","submitted_at":"2026-05-04T15:11:36Z","cross_cats_sorted":["cs.LG","stat.ML"],"title_canon_sha256":"8d9580df019da6feb4704f296d90c73bb3b3f3d4ba7f32d36c8da0007da41cfc","abstract_canon_sha256":"81483593c864179e7b4f2c20037d60301afd256c002c88ce94643f81ab0195c7"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-06-24T01:15:03.316867Z","signature_b64":"rdGEBgqT97Mvv46qMwoCP4PVclpiyRmeqcUfiypJ+ZDZb4U/dZpq3TPmATdJMH3KNGNfn1ssuDIr0ddb+9V9Cw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"3bf3a9e17442207b9e26e16e87261767bc066827e81d9a2eca60b588c355868c","last_reissued_at":"2026-06-24T01:15:03.316448Z","signature_status":"signed_v1","first_computed_at":"2026-06-24T01:15:03.316448Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Robust and Fast Training via Per-Sample Clipping","license":"http://creativecommons.org/licenses/by/4.0/","headline":"Per-sample gradient clipping in SGD achieves optimal convergence rates for non-convex problems under heavy-tailed noise.","cross_cats":["cs.LG","stat.ML"],"primary_cat":"math.OC","authors_text":"Davide Nobile, Philipp Grohs","submitted_at":"2026-05-04T15:11:36Z","abstract_excerpt":"We propose a robust gradient estimator based on per-sample gradient clipping and analyze its properties both theoretically and empirically. We show that the resulting method, per-sample clipped SGD (PS-Clip-SGD), achieves optimal in-expectation convergence rates for non-convex optimization problems under heavy-tailed gradient noise. Moreover, we establish high-probability convergence guarantees that match the in-expectation rates up to polylogarithmic factors in the failure probability. We complement our theoretical results with multiple numerical experiments. In particular, we demonstrate tha"},"claims":{"count":4,"items":[{"kind":"strongest_claim","text":"We show that the resulting method, per-sample clipped SGD (PS-Clip-SGD), achieves optimal in-expectation convergence rates for non-convex optimization problems under heavy-tailed gradient noise. Moreover, we establish high-probability convergence guarantees that match the in-expectation rates up to polylogarithmic factors in the failure probability.","source":"verdict.strongest_claim","status":"machine_extracted","claim_id":"C1","attestation":"unclaimed"},{"kind":"weakest_assumption","text":"The analysis requires that gradient noise follows a heavy-tailed distribution with specific moment bounds; if real gradients during training do not satisfy these tail conditions, the claimed optimal rates may not apply.","source":"verdict.weakest_assumption","status":"machine_extracted","claim_id":"C2","attestation":"unclaimed"},{"kind":"one_line_summary","text":"Per-sample clipped SGD achieves optimal in-expectation and high-probability convergence rates for non-convex optimization under heavy-tailed gradient noise while outperforming standard SGD and batch clipping on CIFAR-100.","source":"verdict.one_line_summary","status":"machine_extracted","claim_id":"C3","attestation":"unclaimed"},{"kind":"headline","text":"Per-sample gradient clipping in SGD achieves optimal convergence rates for non-convex problems under heavy-tailed noise.","source":"verdict.pith_extraction.headline","status":"machine_extracted","claim_id":"C4","attestation":"unclaimed"}],"snapshot_sha256":"96ebf94af9d54dbe16776cd7957191b4554cf98e65ef9e822448b55c7138c5e8"},"source":{"id":"2605.02701","kind":"arxiv","version":2},"verdict":{"id":"b8af9bf2-6242-46d1-9268-5d799912f7e7","model_set":{"reader":"grok-4.3"},"created_at":"2026-05-08T17:46:07.678858Z","strongest_claim":"We show that the resulting method, per-sample clipped SGD (PS-Clip-SGD), achieves optimal in-expectation convergence rates for non-convex optimization problems under heavy-tailed gradient noise. Moreover, we establish high-probability convergence guarantees that match the in-expectation rates up to polylogarithmic factors in the failure probability.","one_line_summary":"Per-sample clipped SGD achieves optimal in-expectation and high-probability convergence rates for non-convex optimization under heavy-tailed gradient noise while outperforming standard SGD and batch clipping on CIFAR-100.","pipeline_version":"pith-pipeline@v0.9.0","weakest_assumption":"The analysis requires that gradient noise follows a heavy-tailed distribution with specific moment bounds; if real gradients during training do not satisfy these tail conditions, the claimed optimal rates may not apply.","pith_extraction_headline":"Per-sample gradient clipping in SGD achieves optimal convergence rates for non-convex problems under heavy-tailed noise."},"integrity":{"clean":true,"summary":{"advisory":0,"critical":0,"by_detector":{},"informational":0},"endpoint":"/pith/2605.02701/integrity.json","findings":[],"available":true,"detectors_run":[{"name":"ai_meta_artifact","ran_at":"2026-05-20T15:35:00.150278Z","status":"completed","version":"1.0.0","findings_count":0},{"name":"doi_title_agreement","ran_at":"2026-05-20T02:31:22.692424Z","status":"completed","version":"1.0.0","findings_count":0},{"name":"doi_compliance","ran_at":"2026-05-19T16:05:45.875652Z","status":"completed","version":"1.0.0","findings_count":0}],"snapshot_sha256":"3e8a2d53656118744d217047f3ce22c3962004c09c874802f25de159473b9e82"},"references":{"count":0,"sample":[],"resolved_work":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57","internal_anchors":0},"formal_canon":{"evidence_count":2,"snapshot_sha256":"38d71a1d2f870b1a569ceea4324294941a6ab30a70be072ab9b090cfdf42ae43"},"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":"2605.02701","created_at":"2026-06-24T01:15:03.316501+00:00"},{"alias_kind":"arxiv_version","alias_value":"2605.02701v2","created_at":"2026-06-24T01:15:03.316501+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2605.02701","created_at":"2026-06-24T01:15:03.316501+00:00"},{"alias_kind":"pith_short_12","alias_value":"HPZ2TYLUIIQH","created_at":"2026-06-24T01:15:03.316501+00:00"},{"alias_kind":"pith_short_16","alias_value":"HPZ2TYLUIIQHXHRG","created_at":"2026-06-24T01:15:03.316501+00:00"},{"alias_kind":"pith_short_8","alias_value":"HPZ2TYLU","created_at":"2026-06-24T01:15:03.316501+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":0,"internal_anchor_count":0,"sample":[]},"formal_canon":{"evidence_count":2,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/HPZ2TYLUIIQHXHRG4FXIOJQXM6","json":"https://pith.science/pith/HPZ2TYLUIIQHXHRG4FXIOJQXM6.json","graph_json":"https://pith.science/api/pith-number/HPZ2TYLUIIQHXHRG4FXIOJQXM6/graph.json","events_json":"https://pith.science/api/pith-number/HPZ2TYLUIIQHXHRG4FXIOJQXM6/events.json","paper":"https://pith.science/paper/HPZ2TYLU"},"agent_actions":{"view_html":"https://pith.science/pith/HPZ2TYLUIIQHXHRG4FXIOJQXM6","download_json":"https://pith.science/pith/HPZ2TYLUIIQHXHRG4FXIOJQXM6.json","view_paper":"https://pith.science/paper/HPZ2TYLU","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2605.02701&json=true","fetch_graph":"https://pith.science/api/pith-number/HPZ2TYLUIIQHXHRG4FXIOJQXM6/graph.json","fetch_events":"https://pith.science/api/pith-number/HPZ2TYLUIIQHXHRG4FXIOJQXM6/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/HPZ2TYLUIIQHXHRG4FXIOJQXM6/action/timestamp_anchor","attest_storage":"https://pith.science/pith/HPZ2TYLUIIQHXHRG4FXIOJQXM6/action/storage_attestation","attest_author":"https://pith.science/pith/HPZ2TYLUIIQHXHRG4FXIOJQXM6/action/author_attestation","sign_citation":"https://pith.science/pith/HPZ2TYLUIIQHXHRG4FXIOJQXM6/action/citation_signature","submit_replication":"https://pith.science/pith/HPZ2TYLUIIQHXHRG4FXIOJQXM6/action/replication_record"}},"created_at":"2026-06-24T01:15:03.316501+00:00","updated_at":"2026-06-24T01:15:03.316501+00:00"}