{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:B2OPJMLCFQ6RMMYXN5PFX5CACY","short_pith_number":"pith:B2OPJMLC","schema_version":"1.0","canonical_sha256":"0e9cf4b1622c3d1633176f5e5bf44016306f8aeb15a339639799ae9af2048adf","source":{"kind":"arxiv","id":"2501.12666","version":1},"attestation_state":"computed","paper":{"title":"Explicit Eigenvalue Regularization Improves Sharpness-Aware Minimization","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.CV"],"primary_cat":"cs.LG","authors_text":"Dinh Phung, Haocheng Luo, Mehrtash Harandi, Trung Le, Tuan Truong, Tung Pham","submitted_at":"2025-01-22T06:03:16Z","abstract_excerpt":"Sharpness-Aware Minimization (SAM) has attracted significant attention for its effectiveness in improving generalization across various tasks. However, its underlying principles remain poorly understood. In this work, we analyze SAM's training dynamics using the maximum eigenvalue of the Hessian as a measure of sharpness, and propose a third-order stochastic differential equation (SDE), which reveals that the dynamics are driven by a complex mixture of second- and third-order terms. We show that alignment between the perturbation vector and the top eigenvector is crucial for SAM's effectivenes"},"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":"2501.12666","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2025-01-22T06:03:16Z","cross_cats_sorted":["cs.CV"],"title_canon_sha256":"20d42e2190be4ba6d5a83497ccbf5c3b8a691f4b15403e65af4c14d93346ab31","abstract_canon_sha256":"139a4b59e27e8cc6d982fd12cc34f3c3715ff60e17dc37125e4e5bb887158ddf"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T10:03:56.732857Z","signature_b64":"tm/tKECf2muwOvOLbZwBL/wRGFk/HCKXMPXJmOVgw7eZ5cP4DRye6ysayjqBoTa/IFs+6WtQj+eByeom+PvFBw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"0e9cf4b1622c3d1633176f5e5bf44016306f8aeb15a339639799ae9af2048adf","last_reissued_at":"2026-07-05T10:03:56.732383Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T10:03:56.732383Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Explicit Eigenvalue Regularization Improves Sharpness-Aware Minimization","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.CV"],"primary_cat":"cs.LG","authors_text":"Dinh Phung, Haocheng Luo, Mehrtash Harandi, Trung Le, Tuan Truong, Tung Pham","submitted_at":"2025-01-22T06:03:16Z","abstract_excerpt":"Sharpness-Aware Minimization (SAM) has attracted significant attention for its effectiveness in improving generalization across various tasks. However, its underlying principles remain poorly understood. In this work, we analyze SAM's training dynamics using the maximum eigenvalue of the Hessian as a measure of sharpness, and propose a third-order stochastic differential equation (SDE), which reveals that the dynamics are driven by a complex mixture of second- and third-order terms. We show that alignment between the perturbation vector and the top eigenvector is crucial for SAM's effectivenes"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2501.12666","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/2501.12666/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":"2501.12666","created_at":"2026-07-05T10:03:56.732451+00:00"},{"alias_kind":"arxiv_version","alias_value":"2501.12666v1","created_at":"2026-07-05T10:03:56.732451+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2501.12666","created_at":"2026-07-05T10:03:56.732451+00:00"},{"alias_kind":"pith_short_12","alias_value":"B2OPJMLCFQ6R","created_at":"2026-07-05T10:03:56.732451+00:00"},{"alias_kind":"pith_short_16","alias_value":"B2OPJMLCFQ6RMMYX","created_at":"2026-07-05T10:03:56.732451+00:00"},{"alias_kind":"pith_short_8","alias_value":"B2OPJMLC","created_at":"2026-07-05T10:03:56.732451+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/B2OPJMLCFQ6RMMYXN5PFX5CACY","json":"https://pith.science/pith/B2OPJMLCFQ6RMMYXN5PFX5CACY.json","graph_json":"https://pith.science/api/pith-number/B2OPJMLCFQ6RMMYXN5PFX5CACY/graph.json","events_json":"https://pith.science/api/pith-number/B2OPJMLCFQ6RMMYXN5PFX5CACY/events.json","paper":"https://pith.science/paper/B2OPJMLC"},"agent_actions":{"view_html":"https://pith.science/pith/B2OPJMLCFQ6RMMYXN5PFX5CACY","download_json":"https://pith.science/pith/B2OPJMLCFQ6RMMYXN5PFX5CACY.json","view_paper":"https://pith.science/paper/B2OPJMLC","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2501.12666&json=true","fetch_graph":"https://pith.science/api/pith-number/B2OPJMLCFQ6RMMYXN5PFX5CACY/graph.json","fetch_events":"https://pith.science/api/pith-number/B2OPJMLCFQ6RMMYXN5PFX5CACY/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/B2OPJMLCFQ6RMMYXN5PFX5CACY/action/timestamp_anchor","attest_storage":"https://pith.science/pith/B2OPJMLCFQ6RMMYXN5PFX5CACY/action/storage_attestation","attest_author":"https://pith.science/pith/B2OPJMLCFQ6RMMYXN5PFX5CACY/action/author_attestation","sign_citation":"https://pith.science/pith/B2OPJMLCFQ6RMMYXN5PFX5CACY/action/citation_signature","submit_replication":"https://pith.science/pith/B2OPJMLCFQ6RMMYXN5PFX5CACY/action/replication_record"}},"created_at":"2026-07-05T10:03:56.732451+00:00","updated_at":"2026-07-05T10:03:56.732451+00:00"}