{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:HNBTOLCFWROJMUVZOE4UL5ZWAQ","short_pith_number":"pith:HNBTOLCF","schema_version":"1.0","canonical_sha256":"3b43372c45b45c9652b9713945f7360402590ce395e9b27d90167a4dca43d91e","source":{"kind":"arxiv","id":"2405.16696","version":2},"attestation_state":"computed","paper":{"title":"How many samples are needed to train a deep neural network?","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["stat.ML","stat.TH"],"primary_cat":"math.ST","authors_text":"Johannes Lederer, Mahsa Taheri, Pegah Golestaneh","submitted_at":"2024-05-26T21:07:21Z","abstract_excerpt":"Neural networks have become standard tools in many areas, yet many important statistical questions remain open. This paper studies the question of how much data are needed to train a ReLU feed-forward neural network. Our theoretical and empirical results suggest that the generalization error of ReLU feed-forward neural networks scales at the rate $1/\\sqrt{n}$ in the sample size $n$ rather than the usual \"parametric rate\" $1/n$. Thus, broadly speaking, our results underpin the common belief that neural networks need \"many\" training samples."},"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":"2405.16696","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"math.ST","submitted_at":"2024-05-26T21:07:21Z","cross_cats_sorted":["stat.ML","stat.TH"],"title_canon_sha256":"6ac8a0e49cac0d3aa9c32fbf6339abb0c0ae5f4a49e8cd11a894dcc73d301e54","abstract_canon_sha256":"01d7f268f740749fc8535411f842d33b2f1c6a295cb84cb12ece8e40209ea231"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:59:05.344676Z","signature_b64":"GanT0BKjWMo59ElxsFtMk86pJFIohrJwke4qmvUV83Xc4uCCXA8/6vaYwRPN//pf/vs7RCjoiAqUiQaJyfAvCA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"3b43372c45b45c9652b9713945f7360402590ce395e9b27d90167a4dca43d91e","last_reissued_at":"2026-07-05T11:59:05.344198Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:59:05.344198Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"How many samples are needed to train a deep neural network?","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["stat.ML","stat.TH"],"primary_cat":"math.ST","authors_text":"Johannes Lederer, Mahsa Taheri, Pegah Golestaneh","submitted_at":"2024-05-26T21:07:21Z","abstract_excerpt":"Neural networks have become standard tools in many areas, yet many important statistical questions remain open. This paper studies the question of how much data are needed to train a ReLU feed-forward neural network. Our theoretical and empirical results suggest that the generalization error of ReLU feed-forward neural networks scales at the rate $1/\\sqrt{n}$ in the sample size $n$ rather than the usual \"parametric rate\" $1/n$. Thus, broadly speaking, our results underpin the common belief that neural networks need \"many\" training samples."},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2405.16696","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/2405.16696/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":"2405.16696","created_at":"2026-07-05T11:59:05.344256+00:00"},{"alias_kind":"arxiv_version","alias_value":"2405.16696v2","created_at":"2026-07-05T11:59:05.344256+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2405.16696","created_at":"2026-07-05T11:59:05.344256+00:00"},{"alias_kind":"pith_short_12","alias_value":"HNBTOLCFWROJ","created_at":"2026-07-05T11:59:05.344256+00:00"},{"alias_kind":"pith_short_16","alias_value":"HNBTOLCFWROJMUVZ","created_at":"2026-07-05T11:59:05.344256+00:00"},{"alias_kind":"pith_short_8","alias_value":"HNBTOLCF","created_at":"2026-07-05T11:59:05.344256+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":2,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.11570","citing_title":"Enhancing Spectral Embedding through Robust and Flexible Knowledge Transfer in Electronic Health Records","ref_index":111,"is_internal_anchor":false},{"citing_arxiv_id":"2605.23583","citing_title":"How Many Training Samples Are Needed for the Inverse Kinematics Solutions by Artificial Neural Networks","ref_index":5,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/HNBTOLCFWROJMUVZOE4UL5ZWAQ","json":"https://pith.science/pith/HNBTOLCFWROJMUVZOE4UL5ZWAQ.json","graph_json":"https://pith.science/api/pith-number/HNBTOLCFWROJMUVZOE4UL5ZWAQ/graph.json","events_json":"https://pith.science/api/pith-number/HNBTOLCFWROJMUVZOE4UL5ZWAQ/events.json","paper":"https://pith.science/paper/HNBTOLCF"},"agent_actions":{"view_html":"https://pith.science/pith/HNBTOLCFWROJMUVZOE4UL5ZWAQ","download_json":"https://pith.science/pith/HNBTOLCFWROJMUVZOE4UL5ZWAQ.json","view_paper":"https://pith.science/paper/HNBTOLCF","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2405.16696&json=true","fetch_graph":"https://pith.science/api/pith-number/HNBTOLCFWROJMUVZOE4UL5ZWAQ/graph.json","fetch_events":"https://pith.science/api/pith-number/HNBTOLCFWROJMUVZOE4UL5ZWAQ/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/HNBTOLCFWROJMUVZOE4UL5ZWAQ/action/timestamp_anchor","attest_storage":"https://pith.science/pith/HNBTOLCFWROJMUVZOE4UL5ZWAQ/action/storage_attestation","attest_author":"https://pith.science/pith/HNBTOLCFWROJMUVZOE4UL5ZWAQ/action/author_attestation","sign_citation":"https://pith.science/pith/HNBTOLCFWROJMUVZOE4UL5ZWAQ/action/citation_signature","submit_replication":"https://pith.science/pith/HNBTOLCFWROJMUVZOE4UL5ZWAQ/action/replication_record"}},"created_at":"2026-07-05T11:59:05.344256+00:00","updated_at":"2026-07-05T11:59:05.344256+00:00"}