{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:J2EDCTECMPB4U6Q7JN4RRRAUCL","short_pith_number":"pith:J2EDCTEC","schema_version":"1.0","canonical_sha256":"4e88314c8263c3ca7a1f4b7918c41412eb064a9326f6517432be2247833a153c","source":{"kind":"arxiv","id":"2505.22578","version":2},"attestation_state":"computed","paper":{"title":"Favorability of Loss Landscape with Weight Decay Requires Both Large Overparametrization and Initialization","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Etienne Boursier, Matthew Bowditch, Matthias Englert, Ranko Lazic","submitted_at":"2025-05-28T16:53:48Z","abstract_excerpt":"The optimization of neural networks under weight decay remains poorly understood from a theoretical standpoint. While weight decay is standard practice in modern training procedures, most theoretical analyses focus on unregularized settings. In this work, we investigate the loss landscape of the $\\ell_2$-regularized training loss for two-layer ReLU networks. We show that the landscape becomes benign -- i.e., free of spurious local minima -- under large overparametrization, specifically when the network width $m$ satisfies $m \\gtrsim \\min(n^d, 2^n)$, where $n$ is the number of data points and $"},"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":"2505.22578","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2025-05-28T16:53:48Z","cross_cats_sorted":[],"title_canon_sha256":"d280e930843e296359b257238ff1e364a744a5c94a42922917773fbd6b39d288","abstract_canon_sha256":"f8d4f0e978d4f2dd8359407864d4edeae3d18474de2ff9af58cf4f7deb874133"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-06-30T02:16:51.967019Z","signature_b64":"ms9aGWfH9Ha98Ijrm+mktpWDSVrIXg3dy3h2XEM4chhh0pqr0s50IitM/tIS0fKDCp7AxZ+0YPoTIYQiK+lKBA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"4e88314c8263c3ca7a1f4b7918c41412eb064a9326f6517432be2247833a153c","last_reissued_at":"2026-06-30T02:16:51.966160Z","signature_status":"signed_v1","first_computed_at":"2026-06-30T02:16:51.966160Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Favorability of Loss Landscape with Weight Decay Requires Both Large Overparametrization and Initialization","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Etienne Boursier, Matthew Bowditch, Matthias Englert, Ranko Lazic","submitted_at":"2025-05-28T16:53:48Z","abstract_excerpt":"The optimization of neural networks under weight decay remains poorly understood from a theoretical standpoint. While weight decay is standard practice in modern training procedures, most theoretical analyses focus on unregularized settings. In this work, we investigate the loss landscape of the $\\ell_2$-regularized training loss for two-layer ReLU networks. We show that the landscape becomes benign -- i.e., free of spurious local minima -- under large overparametrization, specifically when the network width $m$ satisfies $m \\gtrsim \\min(n^d, 2^n)$, where $n$ is the number of data points and $"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2505.22578","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/2505.22578/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":"2505.22578","created_at":"2026-06-30T02:16:51.966261+00:00"},{"alias_kind":"arxiv_version","alias_value":"2505.22578v2","created_at":"2026-06-30T02:16:51.966261+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2505.22578","created_at":"2026-06-30T02:16:51.966261+00:00"},{"alias_kind":"pith_short_12","alias_value":"J2EDCTECMPB4","created_at":"2026-06-30T02:16:51.966261+00:00"},{"alias_kind":"pith_short_16","alias_value":"J2EDCTECMPB4U6Q7","created_at":"2026-06-30T02:16:51.966261+00:00"},{"alias_kind":"pith_short_8","alias_value":"J2EDCTEC","created_at":"2026-06-30T02:16:51.966261+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/J2EDCTECMPB4U6Q7JN4RRRAUCL","json":"https://pith.science/pith/J2EDCTECMPB4U6Q7JN4RRRAUCL.json","graph_json":"https://pith.science/api/pith-number/J2EDCTECMPB4U6Q7JN4RRRAUCL/graph.json","events_json":"https://pith.science/api/pith-number/J2EDCTECMPB4U6Q7JN4RRRAUCL/events.json","paper":"https://pith.science/paper/J2EDCTEC"},"agent_actions":{"view_html":"https://pith.science/pith/J2EDCTECMPB4U6Q7JN4RRRAUCL","download_json":"https://pith.science/pith/J2EDCTECMPB4U6Q7JN4RRRAUCL.json","view_paper":"https://pith.science/paper/J2EDCTEC","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2505.22578&json=true","fetch_graph":"https://pith.science/api/pith-number/J2EDCTECMPB4U6Q7JN4RRRAUCL/graph.json","fetch_events":"https://pith.science/api/pith-number/J2EDCTECMPB4U6Q7JN4RRRAUCL/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/J2EDCTECMPB4U6Q7JN4RRRAUCL/action/timestamp_anchor","attest_storage":"https://pith.science/pith/J2EDCTECMPB4U6Q7JN4RRRAUCL/action/storage_attestation","attest_author":"https://pith.science/pith/J2EDCTECMPB4U6Q7JN4RRRAUCL/action/author_attestation","sign_citation":"https://pith.science/pith/J2EDCTECMPB4U6Q7JN4RRRAUCL/action/citation_signature","submit_replication":"https://pith.science/pith/J2EDCTECMPB4U6Q7JN4RRRAUCL/action/replication_record"}},"created_at":"2026-06-30T02:16:51.966261+00:00","updated_at":"2026-06-30T02:16:51.966261+00:00"}