{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:D2F5LNVA5IAEGWKHICW3NK5XZ6","short_pith_number":"pith:D2F5LNVA","schema_version":"1.0","canonical_sha256":"1e8bd5b6a0ea0043594740adb6abb7cfb753cf8d548402279371e7cf52c12927","source":{"kind":"arxiv","id":"2407.02242","version":2},"attestation_state":"computed","paper":{"title":"Towards optimal hierarchical training of neural networks","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.NA"],"primary_cat":"math.NA","authors_text":"Alexander Rieder, Fabian Zehetgruber, Michael Feischl","submitted_at":"2024-07-02T13:03:46Z","abstract_excerpt":"We propose a hierarchical training algorithm for standard feed-forward neural networks that adaptively extends the network architecture as soon as the optimization reaches a stationary point. By solving small (low-dimensional) optimization problems, the extended network provably escapes any local minimum or stationary point. Under some assumptions on the approximability of the data with stable neural networks, we show that the algorithm achieves an optimal convergence rate s in the sense that loss is bounded by the number of parameters to the -s. As a byproduct, we obtain computable indicators"},"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":"2407.02242","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"math.NA","submitted_at":"2024-07-02T13:03:46Z","cross_cats_sorted":["cs.NA"],"title_canon_sha256":"093a71f2cca055fdd07f295e2e94b12fbe124795d3582b718b3bae3217b94ccc","abstract_canon_sha256":"f1897ba65692d149649e19227519fe6d1b1ce2d22d3c6c487711b1e8706386ea"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T09:28:33.149761Z","signature_b64":"/zf1VEGyX4DwGol3w/ObZZxj+Lz6nNMLo+1HPx3WpmDL8Hz0nmTKVRYZg8VfyeyNNm5FlfS7T9NGKSf4MdlDBw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"1e8bd5b6a0ea0043594740adb6abb7cfb753cf8d548402279371e7cf52c12927","last_reissued_at":"2026-07-05T09:28:33.149247Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T09:28:33.149247Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Towards optimal hierarchical training of neural networks","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.NA"],"primary_cat":"math.NA","authors_text":"Alexander Rieder, Fabian Zehetgruber, Michael Feischl","submitted_at":"2024-07-02T13:03:46Z","abstract_excerpt":"We propose a hierarchical training algorithm for standard feed-forward neural networks that adaptively extends the network architecture as soon as the optimization reaches a stationary point. By solving small (low-dimensional) optimization problems, the extended network provably escapes any local minimum or stationary point. Under some assumptions on the approximability of the data with stable neural networks, we show that the algorithm achieves an optimal convergence rate s in the sense that loss is bounded by the number of parameters to the -s. As a byproduct, we obtain computable indicators"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2407.02242","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/2407.02242/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":"2407.02242","created_at":"2026-07-05T09:28:33.149293+00:00"},{"alias_kind":"arxiv_version","alias_value":"2407.02242v2","created_at":"2026-07-05T09:28:33.149293+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2407.02242","created_at":"2026-07-05T09:28:33.149293+00:00"},{"alias_kind":"pith_short_12","alias_value":"D2F5LNVA5IAE","created_at":"2026-07-05T09:28:33.149293+00:00"},{"alias_kind":"pith_short_16","alias_value":"D2F5LNVA5IAEGWKH","created_at":"2026-07-05T09:28:33.149293+00:00"},{"alias_kind":"pith_short_8","alias_value":"D2F5LNVA","created_at":"2026-07-05T09:28:33.149293+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":1,"sample":[{"citing_arxiv_id":"2505.17751","citing_title":"Computational Math with Neural Networks is Hard","ref_index":2024,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/D2F5LNVA5IAEGWKHICW3NK5XZ6","json":"https://pith.science/pith/D2F5LNVA5IAEGWKHICW3NK5XZ6.json","graph_json":"https://pith.science/api/pith-number/D2F5LNVA5IAEGWKHICW3NK5XZ6/graph.json","events_json":"https://pith.science/api/pith-number/D2F5LNVA5IAEGWKHICW3NK5XZ6/events.json","paper":"https://pith.science/paper/D2F5LNVA"},"agent_actions":{"view_html":"https://pith.science/pith/D2F5LNVA5IAEGWKHICW3NK5XZ6","download_json":"https://pith.science/pith/D2F5LNVA5IAEGWKHICW3NK5XZ6.json","view_paper":"https://pith.science/paper/D2F5LNVA","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2407.02242&json=true","fetch_graph":"https://pith.science/api/pith-number/D2F5LNVA5IAEGWKHICW3NK5XZ6/graph.json","fetch_events":"https://pith.science/api/pith-number/D2F5LNVA5IAEGWKHICW3NK5XZ6/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/D2F5LNVA5IAEGWKHICW3NK5XZ6/action/timestamp_anchor","attest_storage":"https://pith.science/pith/D2F5LNVA5IAEGWKHICW3NK5XZ6/action/storage_attestation","attest_author":"https://pith.science/pith/D2F5LNVA5IAEGWKHICW3NK5XZ6/action/author_attestation","sign_citation":"https://pith.science/pith/D2F5LNVA5IAEGWKHICW3NK5XZ6/action/citation_signature","submit_replication":"https://pith.science/pith/D2F5LNVA5IAEGWKHICW3NK5XZ6/action/replication_record"}},"created_at":"2026-07-05T09:28:33.149293+00:00","updated_at":"2026-07-05T09:28:33.149293+00:00"}