{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2018:Z7MQS5QZZMZDSLFYCPFZ7LAIHA","short_pith_number":"pith:Z7MQS5QZ","schema_version":"1.0","canonical_sha256":"cfd9097619cb32392cb813cb9fac08380f3af91f6b9da7cf70664e5b99a1f811","source":{"kind":"arxiv","id":"1810.05369","version":4},"attestation_state":"computed","paper":{"title":"Regularization Matters: Generalization and Optimization of Neural Nets v.s. their Induced Kernel","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"stat.ML","authors_text":"Colin Wei, Jason D. Lee, Qiang Liu, Tengyu Ma","submitted_at":"2018-10-12T06:21:22Z","abstract_excerpt":"Recent works have shown that on sufficiently over-parametrized neural nets, gradient descent with relatively large initialization optimizes a prediction function in the RKHS of the Neural Tangent Kernel (NTK). This analysis leads to global convergence results but does not work when there is a standard $\\ell_2$ regularizer, which is useful to have in practice. We show that sample efficiency can indeed depend on the presence of the regularizer: we construct a simple distribution in d dimensions which the optimal regularized neural net learns with $O(d)$ samples but the NTK requires $\\Omega(d^2)$"},"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":"1810.05369","kind":"arxiv","version":4},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"stat.ML","submitted_at":"2018-10-12T06:21:22Z","cross_cats_sorted":["cs.LG"],"title_canon_sha256":"d71d4f42a861f62e0a96319294310577dd5e98f221d66e866053a8f4728b7011","abstract_canon_sha256":"0b746898fb740f2dc5fbd40e153813327748e1ac39a0ae6a069c6a1a5c8338b6"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T00:58:01.302793Z","signature_b64":"Sp04mn7sPfK05a7xmvJMBqgM8MthUR7MSGwSbZ53ocIr6fb6xyQHOQgnrgpetXNhZdLa0z+fxBfIkI5sXT5kAA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"cfd9097619cb32392cb813cb9fac08380f3af91f6b9da7cf70664e5b99a1f811","last_reissued_at":"2026-07-05T00:58:01.302200Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T00:58:01.302200Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Regularization Matters: Generalization and Optimization of Neural Nets v.s. their Induced Kernel","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"stat.ML","authors_text":"Colin Wei, Jason D. Lee, Qiang Liu, Tengyu Ma","submitted_at":"2018-10-12T06:21:22Z","abstract_excerpt":"Recent works have shown that on sufficiently over-parametrized neural nets, gradient descent with relatively large initialization optimizes a prediction function in the RKHS of the Neural Tangent Kernel (NTK). This analysis leads to global convergence results but does not work when there is a standard $\\ell_2$ regularizer, which is useful to have in practice. We show that sample efficiency can indeed depend on the presence of the regularizer: we construct a simple distribution in d dimensions which the optimal regularized neural net learns with $O(d)$ samples but the NTK requires $\\Omega(d^2)$"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"1810.05369","kind":"arxiv","version":4},"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/1810.05369/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":"1810.05369","created_at":"2026-07-05T00:58:01.302272+00:00"},{"alias_kind":"arxiv_version","alias_value":"1810.05369v4","created_at":"2026-07-05T00:58:01.302272+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.1810.05369","created_at":"2026-07-05T00:58:01.302272+00:00"},{"alias_kind":"pith_short_12","alias_value":"Z7MQS5QZZMZD","created_at":"2026-07-05T00:58:01.302272+00:00"},{"alias_kind":"pith_short_16","alias_value":"Z7MQS5QZZMZDSLFY","created_at":"2026-07-05T00:58:01.302272+00:00"},{"alias_kind":"pith_short_8","alias_value":"Z7MQS5QZ","created_at":"2026-07-05T00:58:01.302272+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2603.11161","citing_title":"Algorithmic Task Capture, Computational Complexity, and Inductive Bias of Infinite Transformers","ref_index":12,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/Z7MQS5QZZMZDSLFYCPFZ7LAIHA","json":"https://pith.science/pith/Z7MQS5QZZMZDSLFYCPFZ7LAIHA.json","graph_json":"https://pith.science/api/pith-number/Z7MQS5QZZMZDSLFYCPFZ7LAIHA/graph.json","events_json":"https://pith.science/api/pith-number/Z7MQS5QZZMZDSLFYCPFZ7LAIHA/events.json","paper":"https://pith.science/paper/Z7MQS5QZ"},"agent_actions":{"view_html":"https://pith.science/pith/Z7MQS5QZZMZDSLFYCPFZ7LAIHA","download_json":"https://pith.science/pith/Z7MQS5QZZMZDSLFYCPFZ7LAIHA.json","view_paper":"https://pith.science/paper/Z7MQS5QZ","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=1810.05369&json=true","fetch_graph":"https://pith.science/api/pith-number/Z7MQS5QZZMZDSLFYCPFZ7LAIHA/graph.json","fetch_events":"https://pith.science/api/pith-number/Z7MQS5QZZMZDSLFYCPFZ7LAIHA/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/Z7MQS5QZZMZDSLFYCPFZ7LAIHA/action/timestamp_anchor","attest_storage":"https://pith.science/pith/Z7MQS5QZZMZDSLFYCPFZ7LAIHA/action/storage_attestation","attest_author":"https://pith.science/pith/Z7MQS5QZZMZDSLFYCPFZ7LAIHA/action/author_attestation","sign_citation":"https://pith.science/pith/Z7MQS5QZZMZDSLFYCPFZ7LAIHA/action/citation_signature","submit_replication":"https://pith.science/pith/Z7MQS5QZZMZDSLFYCPFZ7LAIHA/action/replication_record"}},"created_at":"2026-07-05T00:58:01.302272+00:00","updated_at":"2026-07-05T00:58:01.302272+00:00"}