{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:HSD5ES37XA7Q6U26TAVH4LTWRL","short_pith_number":"pith:HSD5ES37","schema_version":"1.0","canonical_sha256":"3c87d24b7fb83f0f535e982a7e2e768afaf0921b16d4b7ae26aca95fbbeaa1b1","source":{"kind":"arxiv","id":"2605.29494","version":1},"attestation_state":"computed","paper":{"title":"Gradient Perturbation: Learning to Perturb Gradients for Adaptive Training","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Hua Li","submitted_at":"2026-05-28T07:19:36Z","abstract_excerpt":"Deep neural network training involves both forward propagation (from features through logits to loss) and backward propagation (from loss through gradients to parameter updates). While perturbations along the forward chain, including feature perturbation, logit perturbation, and label perturbation, have been extensively studied, the backward chain's gradient perturbation has received little systematic investigation. In this paper, we establish a unified framework for gradient perturbation, revealing that existing methods such as Sharpness-Aware Minimization (SAM), gradient clipping, and gradie"},"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":"2605.29494","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2026-05-28T07:19:36Z","cross_cats_sorted":[],"title_canon_sha256":"9e4538086537f874e7d2c064f2a806f0a44b6e1db653bf8a595afc5cfe07484d","abstract_canon_sha256":"6e88cb23a9fa2d9c07298ae2968ccb7bb3f422911afd11e7c5cbc2d0959c89c7"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-29T01:05:42.686600Z","signature_b64":"72IRWT68oCfSscwHgz9IUkFgho2aEt0RL1Ti1ZSGxKazrVDt8ahwL/uaPz5kSnwKhfyvClw8m/7JvRwaIPw6Aw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"3c87d24b7fb83f0f535e982a7e2e768afaf0921b16d4b7ae26aca95fbbeaa1b1","last_reissued_at":"2026-05-29T01:05:42.685698Z","signature_status":"signed_v1","first_computed_at":"2026-05-29T01:05:42.685698Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Gradient Perturbation: Learning to Perturb Gradients for Adaptive Training","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Hua Li","submitted_at":"2026-05-28T07:19:36Z","abstract_excerpt":"Deep neural network training involves both forward propagation (from features through logits to loss) and backward propagation (from loss through gradients to parameter updates). While perturbations along the forward chain, including feature perturbation, logit perturbation, and label perturbation, have been extensively studied, the backward chain's gradient perturbation has received little systematic investigation. In this paper, we establish a unified framework for gradient perturbation, revealing that existing methods such as Sharpness-Aware Minimization (SAM), gradient clipping, and gradie"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2605.29494","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/2605.29494/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":"2605.29494","created_at":"2026-05-29T01:05:42.686090+00:00"},{"alias_kind":"arxiv_version","alias_value":"2605.29494v1","created_at":"2026-05-29T01:05:42.686090+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2605.29494","created_at":"2026-05-29T01:05:42.686090+00:00"},{"alias_kind":"pith_short_12","alias_value":"HSD5ES37XA7Q","created_at":"2026-05-29T01:05:42.686090+00:00"},{"alias_kind":"pith_short_16","alias_value":"HSD5ES37XA7Q6U26","created_at":"2026-05-29T01:05:42.686090+00:00"},{"alias_kind":"pith_short_8","alias_value":"HSD5ES37","created_at":"2026-05-29T01:05:42.686090+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":0,"internal_anchor_count":0,"sample":[]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/HSD5ES37XA7Q6U26TAVH4LTWRL","json":"https://pith.science/pith/HSD5ES37XA7Q6U26TAVH4LTWRL.json","graph_json":"https://pith.science/api/pith-number/HSD5ES37XA7Q6U26TAVH4LTWRL/graph.json","events_json":"https://pith.science/api/pith-number/HSD5ES37XA7Q6U26TAVH4LTWRL/events.json","paper":"https://pith.science/paper/HSD5ES37"},"agent_actions":{"view_html":"https://pith.science/pith/HSD5ES37XA7Q6U26TAVH4LTWRL","download_json":"https://pith.science/pith/HSD5ES37XA7Q6U26TAVH4LTWRL.json","view_paper":"https://pith.science/paper/HSD5ES37","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2605.29494&json=true","fetch_graph":"https://pith.science/api/pith-number/HSD5ES37XA7Q6U26TAVH4LTWRL/graph.json","fetch_events":"https://pith.science/api/pith-number/HSD5ES37XA7Q6U26TAVH4LTWRL/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/HSD5ES37XA7Q6U26TAVH4LTWRL/action/timestamp_anchor","attest_storage":"https://pith.science/pith/HSD5ES37XA7Q6U26TAVH4LTWRL/action/storage_attestation","attest_author":"https://pith.science/pith/HSD5ES37XA7Q6U26TAVH4LTWRL/action/author_attestation","sign_citation":"https://pith.science/pith/HSD5ES37XA7Q6U26TAVH4LTWRL/action/citation_signature","submit_replication":"https://pith.science/pith/HSD5ES37XA7Q6U26TAVH4LTWRL/action/replication_record"}},"created_at":"2026-05-29T01:05:42.686090+00:00","updated_at":"2026-05-29T01:05:42.686090+00:00"}