{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:JGKG3E4C4KUB757FGJZMUP2H2A","short_pith_number":"pith:JGKG3E4C","schema_version":"1.0","canonical_sha256":"49946d9382e2a81ff7e53272ca3f47d01efa83ad3739c37becbd388d3925675f","source":{"kind":"arxiv","id":"2605.17767","version":1},"attestation_state":"computed","paper":{"title":"Feature Learning in Linear-Width Two-Layer Networks: Two vs. One Step of Gradient Descent","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"stat.ML","authors_text":"Behrad Moniri, Hamed Hassani","submitted_at":"2026-05-18T02:37:50Z","abstract_excerpt":"We study feature learning in two-layer neural networks within the linear-width regime, where the number of hidden neurons, sample size, and input dimension scale proportionally. While recent work has analyzed feature learning via a single step of gradient descent, such updates are fundamentally limited: they are approximately rank-one, capturing only a single direction, and require the target function to have an information exponent of one. In this paper, we go beyond one-step updates to provide a full characterization of the features learned during the second step of gradient descent with ste"},"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":"2605.17767","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"stat.ML","submitted_at":"2026-05-18T02:37:50Z","cross_cats_sorted":["cs.LG"],"title_canon_sha256":"97cc718f952f42fd07e630d3d3839dd7398a97dc9ab2e785a9cdccc2722bf04e","abstract_canon_sha256":"5ea522cce69429e97441c93055a92e9a212bf281f5a46bd98cf698e95d4939f5"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-20T00:04:57.081848Z","signature_b64":"aLsYPjAsIJQxat7FCHpNzFJ0IhM3GzdEkWv/yIKTywL9NQ7/QB/963fAR+lLPR/RqzocVVp5bXbS1ihX6e+6Ag==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"49946d9382e2a81ff7e53272ca3f47d01efa83ad3739c37becbd388d3925675f","last_reissued_at":"2026-05-20T00:04:57.081043Z","signature_status":"signed_v1","first_computed_at":"2026-05-20T00:04:57.081043Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Feature Learning in Linear-Width Two-Layer Networks: Two vs. One Step of Gradient Descent","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"stat.ML","authors_text":"Behrad Moniri, Hamed Hassani","submitted_at":"2026-05-18T02:37:50Z","abstract_excerpt":"We study feature learning in two-layer neural networks within the linear-width regime, where the number of hidden neurons, sample size, and input dimension scale proportionally. While recent work has analyzed feature learning via a single step of gradient descent, such updates are fundamentally limited: they are approximately rank-one, capturing only a single direction, and require the target function to have an information exponent of one. In this paper, we go beyond one-step updates to provide a full characterization of the features learned during the second step of gradient descent with ste"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2605.17767","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/2605.17767/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":"2605.17767","created_at":"2026-05-20T00:04:57.081184+00:00"},{"alias_kind":"arxiv_version","alias_value":"2605.17767v1","created_at":"2026-05-20T00:04:57.081184+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2605.17767","created_at":"2026-05-20T00:04:57.081184+00:00"},{"alias_kind":"pith_short_12","alias_value":"JGKG3E4C4KUB","created_at":"2026-05-20T00:04:57.081184+00:00"},{"alias_kind":"pith_short_16","alias_value":"JGKG3E4C4KUB757F","created_at":"2026-05-20T00:04:57.081184+00:00"},{"alias_kind":"pith_short_8","alias_value":"JGKG3E4C","created_at":"2026-05-20T00:04:57.081184+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":0,"internal_anchor_count":0,"sample":[]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/JGKG3E4C4KUB757FGJZMUP2H2A","json":"https://pith.science/pith/JGKG3E4C4KUB757FGJZMUP2H2A.json","graph_json":"https://pith.science/api/pith-number/JGKG3E4C4KUB757FGJZMUP2H2A/graph.json","events_json":"https://pith.science/api/pith-number/JGKG3E4C4KUB757FGJZMUP2H2A/events.json","paper":"https://pith.science/paper/JGKG3E4C"},"agent_actions":{"view_html":"https://pith.science/pith/JGKG3E4C4KUB757FGJZMUP2H2A","download_json":"https://pith.science/pith/JGKG3E4C4KUB757FGJZMUP2H2A.json","view_paper":"https://pith.science/paper/JGKG3E4C","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2605.17767&json=true","fetch_graph":"https://pith.science/api/pith-number/JGKG3E4C4KUB757FGJZMUP2H2A/graph.json","fetch_events":"https://pith.science/api/pith-number/JGKG3E4C4KUB757FGJZMUP2H2A/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/JGKG3E4C4KUB757FGJZMUP2H2A/action/timestamp_anchor","attest_storage":"https://pith.science/pith/JGKG3E4C4KUB757FGJZMUP2H2A/action/storage_attestation","attest_author":"https://pith.science/pith/JGKG3E4C4KUB757FGJZMUP2H2A/action/author_attestation","sign_citation":"https://pith.science/pith/JGKG3E4C4KUB757FGJZMUP2H2A/action/citation_signature","submit_replication":"https://pith.science/pith/JGKG3E4C4KUB757FGJZMUP2H2A/action/replication_record"}},"created_at":"2026-05-20T00:04:57.081184+00:00","updated_at":"2026-05-20T00:04:57.081184+00:00"}