{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2018:5YRJPRNPIKUSYIY4VYKAEQKZOU","short_pith_number":"pith:5YRJPRNP","schema_version":"1.0","canonical_sha256":"ee2297c5af42a92c231cae14024159753d576153503bcfe1b039b89545d1e96f","source":{"kind":"arxiv","id":"1810.07770","version":3},"attestation_state":"computed","paper":{"title":"Small ReLU networks are powerful memorizers: a tight analysis of memorization capacity","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["stat.ML"],"primary_cat":"cs.LG","authors_text":"Ali Jadbabaie, Chulhee Yun, Suvrit Sra","submitted_at":"2018-10-17T20:21:43Z","abstract_excerpt":"We study finite sample expressivity, i.e., memorization power of ReLU networks. Recent results require $N$ hidden nodes to memorize/interpolate arbitrary $N$ data points. In contrast, by exploiting depth, we show that 3-layer ReLU networks with $\\Omega(\\sqrt{N})$ hidden nodes can perfectly memorize most datasets with $N$ points. We also prove that width $\\Theta(\\sqrt{N})$ is necessary and sufficient for memorizing $N$ data points, proving tight bounds on memorization capacity. The sufficiency result can be extended to deeper networks; we show that an $L$-layer network with $W$ parameters in th"},"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":"1810.07770","kind":"arxiv","version":3},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2018-10-17T20:21:43Z","cross_cats_sorted":["stat.ML"],"title_canon_sha256":"2d76acba5bd2764cb2a1691183499f67b0423b97be255bafc4cadbe62c6d6deb","abstract_canon_sha256":"ab2d79e91af71e79220ead7e5df5a45811ffc7fe0b714e107d0118db53270449"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T00:15:22.037304Z","signature_b64":"HXMfPexfhnuoNaikFAj59H67yPrJqhcbPZIjHHSJt1vKQRIprPADwuHtrzds/t5OihGDCTK6Yk6g2CU99bnKCQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"ee2297c5af42a92c231cae14024159753d576153503bcfe1b039b89545d1e96f","last_reissued_at":"2026-07-05T00:15:22.036827Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T00:15:22.036827Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Small ReLU networks are powerful memorizers: a tight analysis of memorization capacity","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["stat.ML"],"primary_cat":"cs.LG","authors_text":"Ali Jadbabaie, Chulhee Yun, Suvrit Sra","submitted_at":"2018-10-17T20:21:43Z","abstract_excerpt":"We study finite sample expressivity, i.e., memorization power of ReLU networks. Recent results require $N$ hidden nodes to memorize/interpolate arbitrary $N$ data points. In contrast, by exploiting depth, we show that 3-layer ReLU networks with $\\Omega(\\sqrt{N})$ hidden nodes can perfectly memorize most datasets with $N$ points. We also prove that width $\\Theta(\\sqrt{N})$ is necessary and sufficient for memorizing $N$ data points, proving tight bounds on memorization capacity. The sufficiency result can be extended to deeper networks; we show that an $L$-layer network with $W$ parameters in th"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"1810.07770","kind":"arxiv","version":3},"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/1810.07770/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":"1810.07770","created_at":"2026-07-05T00:15:22.036885+00:00"},{"alias_kind":"arxiv_version","alias_value":"1810.07770v3","created_at":"2026-07-05T00:15:22.036885+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.1810.07770","created_at":"2026-07-05T00:15:22.036885+00:00"},{"alias_kind":"pith_short_12","alias_value":"5YRJPRNPIKUS","created_at":"2026-07-05T00:15:22.036885+00:00"},{"alias_kind":"pith_short_16","alias_value":"5YRJPRNPIKUSYIY4","created_at":"2026-07-05T00:15:22.036885+00:00"},{"alias_kind":"pith_short_8","alias_value":"5YRJPRNP","created_at":"2026-07-05T00:15:22.036885+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2605.24244","citing_title":"MEDAL: Manifold Embedding Distillation via Autoencoder Learning","ref_index":116,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/5YRJPRNPIKUSYIY4VYKAEQKZOU","json":"https://pith.science/pith/5YRJPRNPIKUSYIY4VYKAEQKZOU.json","graph_json":"https://pith.science/api/pith-number/5YRJPRNPIKUSYIY4VYKAEQKZOU/graph.json","events_json":"https://pith.science/api/pith-number/5YRJPRNPIKUSYIY4VYKAEQKZOU/events.json","paper":"https://pith.science/paper/5YRJPRNP"},"agent_actions":{"view_html":"https://pith.science/pith/5YRJPRNPIKUSYIY4VYKAEQKZOU","download_json":"https://pith.science/pith/5YRJPRNPIKUSYIY4VYKAEQKZOU.json","view_paper":"https://pith.science/paper/5YRJPRNP","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=1810.07770&json=true","fetch_graph":"https://pith.science/api/pith-number/5YRJPRNPIKUSYIY4VYKAEQKZOU/graph.json","fetch_events":"https://pith.science/api/pith-number/5YRJPRNPIKUSYIY4VYKAEQKZOU/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/5YRJPRNPIKUSYIY4VYKAEQKZOU/action/timestamp_anchor","attest_storage":"https://pith.science/pith/5YRJPRNPIKUSYIY4VYKAEQKZOU/action/storage_attestation","attest_author":"https://pith.science/pith/5YRJPRNPIKUSYIY4VYKAEQKZOU/action/author_attestation","sign_citation":"https://pith.science/pith/5YRJPRNPIKUSYIY4VYKAEQKZOU/action/citation_signature","submit_replication":"https://pith.science/pith/5YRJPRNPIKUSYIY4VYKAEQKZOU/action/replication_record"}},"created_at":"2026-07-05T00:15:22.036885+00:00","updated_at":"2026-07-05T00:15:22.036885+00:00"}