{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2022:72RIDS4J2BM3DGJ6BPQPXTVYNP","short_pith_number":"pith:72RIDS4J","schema_version":"1.0","canonical_sha256":"fea281cb89d059b1993e0be0fbceb86beb6228b46031da134b4943336a8868d1","source":{"kind":"arxiv","id":"2205.12808","version":2},"attestation_state":"computed","paper":{"title":"Mirror Descent Maximizes Generalized Margin and Can Be Implemented Efficiently","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Christos Thrampoulidis, Haoyuan Sun, Kwangjun Ahn, Navid Azizan","submitted_at":"2022-05-25T14:33:13Z","abstract_excerpt":"Driven by the empirical success and wide use of deep neural networks, understanding the generalization performance of overparameterized models has become an increasingly popular question. To this end, there has been substantial effort to characterize the implicit bias of the optimization algorithms used, such as gradient descent (GD), and the structural properties of their preferred solutions. This paper answers an open question in this literature: For the classification setting, what solution does mirror descent (MD) converge to? Specifically, motivated by its efficient implementation, we con"},"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":"2205.12808","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2022-05-25T14:33:13Z","cross_cats_sorted":[],"title_canon_sha256":"350a88ce1c92486c5bf944d58fb52d90d053af19de6ab512c81f450dfb0f353e","abstract_canon_sha256":"c7f3c2ba83ebf3437ce9eb29e8abe9f6e0065788bde8d7be9df57965067694b0"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T06:24:09.258919Z","signature_b64":"sh/hvYajrKAbu8HH2/NNyHGRb+ryn5KD9di/qGjX2+cln4rMgCaaqAJ89+xTdVd740pf4/mdNQdV3fe88+2/AQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"fea281cb89d059b1993e0be0fbceb86beb6228b46031da134b4943336a8868d1","last_reissued_at":"2026-07-05T06:24:09.258527Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T06:24:09.258527Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Mirror Descent Maximizes Generalized Margin and Can Be Implemented Efficiently","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Christos Thrampoulidis, Haoyuan Sun, Kwangjun Ahn, Navid Azizan","submitted_at":"2022-05-25T14:33:13Z","abstract_excerpt":"Driven by the empirical success and wide use of deep neural networks, understanding the generalization performance of overparameterized models has become an increasingly popular question. To this end, there has been substantial effort to characterize the implicit bias of the optimization algorithms used, such as gradient descent (GD), and the structural properties of their preferred solutions. This paper answers an open question in this literature: For the classification setting, what solution does mirror descent (MD) converge to? Specifically, motivated by its efficient implementation, we con"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2205.12808","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/2205.12808/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":"2205.12808","created_at":"2026-07-05T06:24:09.258583+00:00"},{"alias_kind":"arxiv_version","alias_value":"2205.12808v2","created_at":"2026-07-05T06:24:09.258583+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2205.12808","created_at":"2026-07-05T06:24:09.258583+00:00"},{"alias_kind":"pith_short_12","alias_value":"72RIDS4J2BM3","created_at":"2026-07-05T06:24:09.258583+00:00"},{"alias_kind":"pith_short_16","alias_value":"72RIDS4J2BM3DGJ6","created_at":"2026-07-05T06:24:09.258583+00:00"},{"alias_kind":"pith_short_8","alias_value":"72RIDS4J","created_at":"2026-07-05T06:24:09.258583+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":1,"sample":[{"citing_arxiv_id":"2607.06013","citing_title":"Stability Annealing Selects the Implicit Bias of Smoothed Sign Descent: A Rate-Indexed Barrier Path on Separable Data","ref_index":13,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/72RIDS4J2BM3DGJ6BPQPXTVYNP","json":"https://pith.science/pith/72RIDS4J2BM3DGJ6BPQPXTVYNP.json","graph_json":"https://pith.science/api/pith-number/72RIDS4J2BM3DGJ6BPQPXTVYNP/graph.json","events_json":"https://pith.science/api/pith-number/72RIDS4J2BM3DGJ6BPQPXTVYNP/events.json","paper":"https://pith.science/paper/72RIDS4J"},"agent_actions":{"view_html":"https://pith.science/pith/72RIDS4J2BM3DGJ6BPQPXTVYNP","download_json":"https://pith.science/pith/72RIDS4J2BM3DGJ6BPQPXTVYNP.json","view_paper":"https://pith.science/paper/72RIDS4J","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2205.12808&json=true","fetch_graph":"https://pith.science/api/pith-number/72RIDS4J2BM3DGJ6BPQPXTVYNP/graph.json","fetch_events":"https://pith.science/api/pith-number/72RIDS4J2BM3DGJ6BPQPXTVYNP/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/72RIDS4J2BM3DGJ6BPQPXTVYNP/action/timestamp_anchor","attest_storage":"https://pith.science/pith/72RIDS4J2BM3DGJ6BPQPXTVYNP/action/storage_attestation","attest_author":"https://pith.science/pith/72RIDS4J2BM3DGJ6BPQPXTVYNP/action/author_attestation","sign_citation":"https://pith.science/pith/72RIDS4J2BM3DGJ6BPQPXTVYNP/action/citation_signature","submit_replication":"https://pith.science/pith/72RIDS4J2BM3DGJ6BPQPXTVYNP/action/replication_record"}},"created_at":"2026-07-05T06:24:09.258583+00:00","updated_at":"2026-07-05T06:24:09.258583+00:00"}