{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2022:PQZDIVIS46N54ZK2O7RLVPZQXH","short_pith_number":"pith:PQZDIVIS","schema_version":"1.0","canonical_sha256":"7c32345512e79bde655a77e2babf30b9e332bbe1f7463d6db4633c70d7488e42","source":{"kind":"arxiv","id":"2212.02323","version":1},"attestation_state":"computed","paper":{"title":"Improved Convergence Guarantees for Shallow Neural Networks","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Alexander Razborov","submitted_at":"2022-12-05T14:47:52Z","abstract_excerpt":"We continue a long line of research aimed at proving convergence of depth 2 neural networks, trained via gradient descent, to a global minimum. Like in many previous works, our model has the following features: regression with quadratic loss function, fully connected feedforward architecture, RelU activations, Gaussian data instances and network initialization, adversarial labels. It is more general in the sense that we allow both layers to be trained simultaneously and at {\\em different} rates.\n  Our results improve on state-of-the-art [Oymak Soltanolkotabi 20] (training the first layer only)"},"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":"2212.02323","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2022-12-05T14:47:52Z","cross_cats_sorted":[],"title_canon_sha256":"c7bdbd99488795bbfd5cad273418a66ad0fbca7b5071d69ef804ac0a2cf7f61e","abstract_canon_sha256":"ded821f363f4772ea7932eead4a90316a5f5bcf6843bbb678cf60381b9094d7e"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T05:22:23.094274Z","signature_b64":"I095IvIkbEtuEQYTvMrJojXsRTZkvw0RFZog2gJb+Pm6q2wOLTtO9b0hWO8kjzH36IMBdcNSYiI1hlgqs/3uBA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"7c32345512e79bde655a77e2babf30b9e332bbe1f7463d6db4633c70d7488e42","last_reissued_at":"2026-07-05T05:22:23.093892Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T05:22:23.093892Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Improved Convergence Guarantees for Shallow Neural Networks","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Alexander Razborov","submitted_at":"2022-12-05T14:47:52Z","abstract_excerpt":"We continue a long line of research aimed at proving convergence of depth 2 neural networks, trained via gradient descent, to a global minimum. Like in many previous works, our model has the following features: regression with quadratic loss function, fully connected feedforward architecture, RelU activations, Gaussian data instances and network initialization, adversarial labels. It is more general in the sense that we allow both layers to be trained simultaneously and at {\\em different} rates.\n  Our results improve on state-of-the-art [Oymak Soltanolkotabi 20] (training the first layer only)"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2212.02323","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/2212.02323/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":"2212.02323","created_at":"2026-07-05T05:22:23.093954+00:00"},{"alias_kind":"arxiv_version","alias_value":"2212.02323v1","created_at":"2026-07-05T05:22:23.093954+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2212.02323","created_at":"2026-07-05T05:22:23.093954+00:00"},{"alias_kind":"pith_short_12","alias_value":"PQZDIVIS46N5","created_at":"2026-07-05T05:22:23.093954+00:00"},{"alias_kind":"pith_short_16","alias_value":"PQZDIVIS46N54ZK2","created_at":"2026-07-05T05:22:23.093954+00:00"},{"alias_kind":"pith_short_8","alias_value":"PQZDIVIS","created_at":"2026-07-05T05:22:23.093954+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":1,"sample":[{"citing_arxiv_id":"2505.11621","citing_title":"A Classical View on Benign Overfitting: The Role of Sample Size","ref_index":80,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/PQZDIVIS46N54ZK2O7RLVPZQXH","json":"https://pith.science/pith/PQZDIVIS46N54ZK2O7RLVPZQXH.json","graph_json":"https://pith.science/api/pith-number/PQZDIVIS46N54ZK2O7RLVPZQXH/graph.json","events_json":"https://pith.science/api/pith-number/PQZDIVIS46N54ZK2O7RLVPZQXH/events.json","paper":"https://pith.science/paper/PQZDIVIS"},"agent_actions":{"view_html":"https://pith.science/pith/PQZDIVIS46N54ZK2O7RLVPZQXH","download_json":"https://pith.science/pith/PQZDIVIS46N54ZK2O7RLVPZQXH.json","view_paper":"https://pith.science/paper/PQZDIVIS","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2212.02323&json=true","fetch_graph":"https://pith.science/api/pith-number/PQZDIVIS46N54ZK2O7RLVPZQXH/graph.json","fetch_events":"https://pith.science/api/pith-number/PQZDIVIS46N54ZK2O7RLVPZQXH/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/PQZDIVIS46N54ZK2O7RLVPZQXH/action/timestamp_anchor","attest_storage":"https://pith.science/pith/PQZDIVIS46N54ZK2O7RLVPZQXH/action/storage_attestation","attest_author":"https://pith.science/pith/PQZDIVIS46N54ZK2O7RLVPZQXH/action/author_attestation","sign_citation":"https://pith.science/pith/PQZDIVIS46N54ZK2O7RLVPZQXH/action/citation_signature","submit_replication":"https://pith.science/pith/PQZDIVIS46N54ZK2O7RLVPZQXH/action/replication_record"}},"created_at":"2026-07-05T05:22:23.093954+00:00","updated_at":"2026-07-05T05:22:23.093954+00:00"}