{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2019:LCXWXHK527LASNHNOV7BUSIR42","short_pith_number":"pith:LCXWXHK5","schema_version":"1.0","canonical_sha256":"58af6b9d5dd7d60934ed757e1a4911e6a08119f23b91e1f39ec0433034fa6da8","source":{"kind":"arxiv","id":"1903.02237","version":1},"attestation_state":"computed","paper":{"title":"Positively Scale-Invariant Flatness of ReLU Neural Networks","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["stat.ML"],"primary_cat":"cs.LG","authors_text":"Mingyang Yi, Qi Meng, Tie-Yan Liu, Wei Chen, Zhi-Ming Ma","submitted_at":"2019-03-06T08:21:09Z","abstract_excerpt":"It was empirically confirmed by Keskar et al.\\cite{SharpMinima} that flatter minima generalize better. However, for the popular ReLU network, sharp minimum can also generalize well \\cite{SharpMinimacan}. The conclusion demonstrates that the existing definitions of flatness fail to account for the complex geometry of ReLU neural networks because they can't cover the Positively Scale-Invariant (PSI) property of ReLU network. In this paper, we formalize the PSI causes problem of existing definitions of flatness and propose a new description of flatness - \\emph{PSI-flatness}. PSI-flatness is defin"},"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":"1903.02237","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2019-03-06T08:21:09Z","cross_cats_sorted":["stat.ML"],"title_canon_sha256":"1b701d0b4e57f4de3fb3f38858f3916441197d56882e1c7981be584b5ae5194d","abstract_canon_sha256":"4bdb5e0f220cfbef030e303a0bac6430aae5f781eb01e1bab350da0049cb5da5"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-17T23:51:56.422859Z","signature_b64":"1GdWx7mC8dFcC1RS2x2h0cImnnepzRPFhEmJ0lFPRTVFS9ggg5JNhqXPhzOMWxi1g8IPL/tD3/J2u5Gnza5jAw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"58af6b9d5dd7d60934ed757e1a4911e6a08119f23b91e1f39ec0433034fa6da8","last_reissued_at":"2026-05-17T23:51:56.422410Z","signature_status":"signed_v1","first_computed_at":"2026-05-17T23:51:56.422410Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Positively Scale-Invariant Flatness of ReLU Neural Networks","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["stat.ML"],"primary_cat":"cs.LG","authors_text":"Mingyang Yi, Qi Meng, Tie-Yan Liu, Wei Chen, Zhi-Ming Ma","submitted_at":"2019-03-06T08:21:09Z","abstract_excerpt":"It was empirically confirmed by Keskar et al.\\cite{SharpMinima} that flatter minima generalize better. However, for the popular ReLU network, sharp minimum can also generalize well \\cite{SharpMinimacan}. The conclusion demonstrates that the existing definitions of flatness fail to account for the complex geometry of ReLU neural networks because they can't cover the Positively Scale-Invariant (PSI) property of ReLU network. In this paper, we formalize the PSI causes problem of existing definitions of flatness and propose a new description of flatness - \\emph{PSI-flatness}. PSI-flatness is defin"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"1903.02237","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":""},"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":"1903.02237","created_at":"2026-05-17T23:51:56.422483+00:00"},{"alias_kind":"arxiv_version","alias_value":"1903.02237v1","created_at":"2026-05-17T23:51:56.422483+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.1903.02237","created_at":"2026-05-17T23:51:56.422483+00:00"},{"alias_kind":"pith_short_12","alias_value":"LCXWXHK527LA","created_at":"2026-05-18T12:33:21.387695+00:00"},{"alias_kind":"pith_short_16","alias_value":"LCXWXHK527LASNHN","created_at":"2026-05-18T12:33:21.387695+00:00"},{"alias_kind":"pith_short_8","alias_value":"LCXWXHK5","created_at":"2026-05-18T12:33:21.387695+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":2,"internal_anchor_count":2,"sample":[{"citing_arxiv_id":"1907.10732","citing_title":"Hessian based analysis of SGD for Deep Nets: Dynamics and Generalization","ref_index":78,"is_internal_anchor":true},{"citing_arxiv_id":"2602.01642","citing_title":"The Effect of Mini-Batch Noise on the Implicit Bias of Adam","ref_index":74,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/LCXWXHK527LASNHNOV7BUSIR42","json":"https://pith.science/pith/LCXWXHK527LASNHNOV7BUSIR42.json","graph_json":"https://pith.science/api/pith-number/LCXWXHK527LASNHNOV7BUSIR42/graph.json","events_json":"https://pith.science/api/pith-number/LCXWXHK527LASNHNOV7BUSIR42/events.json","paper":"https://pith.science/paper/LCXWXHK5"},"agent_actions":{"view_html":"https://pith.science/pith/LCXWXHK527LASNHNOV7BUSIR42","download_json":"https://pith.science/pith/LCXWXHK527LASNHNOV7BUSIR42.json","view_paper":"https://pith.science/paper/LCXWXHK5","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=1903.02237&json=true","fetch_graph":"https://pith.science/api/pith-number/LCXWXHK527LASNHNOV7BUSIR42/graph.json","fetch_events":"https://pith.science/api/pith-number/LCXWXHK527LASNHNOV7BUSIR42/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/LCXWXHK527LASNHNOV7BUSIR42/action/timestamp_anchor","attest_storage":"https://pith.science/pith/LCXWXHK527LASNHNOV7BUSIR42/action/storage_attestation","attest_author":"https://pith.science/pith/LCXWXHK527LASNHNOV7BUSIR42/action/author_attestation","sign_citation":"https://pith.science/pith/LCXWXHK527LASNHNOV7BUSIR42/action/citation_signature","submit_replication":"https://pith.science/pith/LCXWXHK527LASNHNOV7BUSIR42/action/replication_record"}},"created_at":"2026-05-17T23:51:56.422483+00:00","updated_at":"2026-05-17T23:51:56.422483+00:00"}