{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:IJW66WYFLJHZMKSQ4MP7HDEGNC","short_pith_number":"pith:IJW66WYF","schema_version":"1.0","canonical_sha256":"426def5b055a4f962a50e31ff38c8668bed85cb2dc87b35bb0e23d9336d640b5","source":{"kind":"arxiv","id":"2405.15459","version":2},"attestation_state":"computed","paper":{"title":"Repetita Iuvant: Data Repetition Allows SGD to Learn High-Dimensional Multi-Index Functions","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"stat.ML","authors_text":"Florent Krzakala, Luca Arnaboldi, Luca Pesce, Ludovic Stephan, Yatin Dandi","submitted_at":"2024-05-24T11:34:31Z","abstract_excerpt":"Neural networks can identify low-dimensional relevant structures within high-dimensional noisy data, yet our mathematical understanding of how they do so remains scarce. Here, we investigate the training dynamics of two-layer shallow neural networks trained with gradient-based algorithms, and discuss how they learn pertinent features in multi-index models, that is target functions with low-dimensional relevant directions. In the high-dimensional regime, where the input dimension $d$ diverges, we show that a simple modification of the idealized single-pass gradient descent training scenario, wh"},"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":"2405.15459","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"stat.ML","submitted_at":"2024-05-24T11:34:31Z","cross_cats_sorted":["cs.LG"],"title_canon_sha256":"85aeedcd43e88f427fc5d3f6d8a95f044336e2f44682da55ad8392a239c11ed8","abstract_canon_sha256":"31989f22ef822b3322c0a90b4a979554aff1b44a4cde9f6e0cfda9cfdf79a17b"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T10:11:19.773458Z","signature_b64":"JwNTRWotdKB7fqKSilyxzEh2Vd9MU3YK9N93SXYUtfAHbhWkAYdazppi/GtvXbplbVh6Bp+j86yV9vfCgLIIAQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"426def5b055a4f962a50e31ff38c8668bed85cb2dc87b35bb0e23d9336d640b5","last_reissued_at":"2026-07-05T10:11:19.772923Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T10:11:19.772923Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Repetita Iuvant: Data Repetition Allows SGD to Learn High-Dimensional Multi-Index Functions","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"stat.ML","authors_text":"Florent Krzakala, Luca Arnaboldi, Luca Pesce, Ludovic Stephan, Yatin Dandi","submitted_at":"2024-05-24T11:34:31Z","abstract_excerpt":"Neural networks can identify low-dimensional relevant structures within high-dimensional noisy data, yet our mathematical understanding of how they do so remains scarce. Here, we investigate the training dynamics of two-layer shallow neural networks trained with gradient-based algorithms, and discuss how they learn pertinent features in multi-index models, that is target functions with low-dimensional relevant directions. In the high-dimensional regime, where the input dimension $d$ diverges, we show that a simple modification of the idealized single-pass gradient descent training scenario, wh"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2405.15459","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/2405.15459/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":"2405.15459","created_at":"2026-07-05T10:11:19.772990+00:00"},{"alias_kind":"arxiv_version","alias_value":"2405.15459v2","created_at":"2026-07-05T10:11:19.772990+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2405.15459","created_at":"2026-07-05T10:11:19.772990+00:00"},{"alias_kind":"pith_short_12","alias_value":"IJW66WYFLJHZ","created_at":"2026-07-05T10:11:19.772990+00:00"},{"alias_kind":"pith_short_16","alias_value":"IJW66WYFLJHZMKSQ","created_at":"2026-07-05T10:11:19.772990+00:00"},{"alias_kind":"pith_short_8","alias_value":"IJW66WYF","created_at":"2026-07-05T10:11:19.772990+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":3,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2605.24749","citing_title":"How Neural Reward Models Learn Features for Policy Optimization: A Single-Index Analysis","ref_index":2,"is_internal_anchor":false},{"citing_arxiv_id":"2605.15082","citing_title":"Average Gradient Outer Product in kernel regression provably recovers the central subspace for multi-index models","ref_index":6,"is_internal_anchor":false},{"citing_arxiv_id":"2605.14567","citing_title":"Scaling Laws from Sequential Feature Recovery: A Solvable Hierarchical Model","ref_index":166,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/IJW66WYFLJHZMKSQ4MP7HDEGNC","json":"https://pith.science/pith/IJW66WYFLJHZMKSQ4MP7HDEGNC.json","graph_json":"https://pith.science/api/pith-number/IJW66WYFLJHZMKSQ4MP7HDEGNC/graph.json","events_json":"https://pith.science/api/pith-number/IJW66WYFLJHZMKSQ4MP7HDEGNC/events.json","paper":"https://pith.science/paper/IJW66WYF"},"agent_actions":{"view_html":"https://pith.science/pith/IJW66WYFLJHZMKSQ4MP7HDEGNC","download_json":"https://pith.science/pith/IJW66WYFLJHZMKSQ4MP7HDEGNC.json","view_paper":"https://pith.science/paper/IJW66WYF","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2405.15459&json=true","fetch_graph":"https://pith.science/api/pith-number/IJW66WYFLJHZMKSQ4MP7HDEGNC/graph.json","fetch_events":"https://pith.science/api/pith-number/IJW66WYFLJHZMKSQ4MP7HDEGNC/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/IJW66WYFLJHZMKSQ4MP7HDEGNC/action/timestamp_anchor","attest_storage":"https://pith.science/pith/IJW66WYFLJHZMKSQ4MP7HDEGNC/action/storage_attestation","attest_author":"https://pith.science/pith/IJW66WYFLJHZMKSQ4MP7HDEGNC/action/author_attestation","sign_citation":"https://pith.science/pith/IJW66WYFLJHZMKSQ4MP7HDEGNC/action/citation_signature","submit_replication":"https://pith.science/pith/IJW66WYFLJHZMKSQ4MP7HDEGNC/action/replication_record"}},"created_at":"2026-07-05T10:11:19.772990+00:00","updated_at":"2026-07-05T10:11:19.772990+00:00"}