{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2021:K4FBDPUN4M2P46USUSJE3O2QFW","short_pith_number":"pith:K4FBDPUN","schema_version":"1.0","canonical_sha256":"570a11be8de334fe7a92a4924dbb502d9b2086628711b4b09f8d6d9c7b9f1f41","source":{"kind":"arxiv","id":"2101.02944","version":1},"attestation_state":"computed","paper":{"title":"BN-invariant sharpness regularizes the training model to better generalization","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Huishuai Zhang, Mingyang Yi, Tie-Yan Liu, Wei Chen, Zhi-Ming Ma","submitted_at":"2021-01-08T10:23:24Z","abstract_excerpt":"It is arguably believed that flatter minima can generalize better. However, it has been pointed out that the usual definitions of sharpness, which consider either the maxima or the integral of loss over a $\\delta$ ball of parameters around minima, cannot give consistent measurement for scale invariant neural networks, e.g., networks with batch normalization layer. In this paper, we first propose a measure of sharpness, BN-Sharpness, which gives consistent value for equivalent networks under BN. It achieves the property of scale invariance by connecting the integral diameter with the scale of p"},"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":"2101.02944","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2021-01-08T10:23:24Z","cross_cats_sorted":[],"title_canon_sha256":"dc98cea1f4c013d0d4383debfba1e812fc38bac6c7cbf3b0902e9dc1518786e5","abstract_canon_sha256":"cc91a7acfe9494d816ee9f35bfa222f3cef244ec6b48b49ddea535fadd833943"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T02:05:32.415240Z","signature_b64":"Vt/RnR75fZH20nonLHtOvA8atc77r200Dk9mTrCNtjvQodsfmO5tbcfPHipML0Z9U2VbjtOY1afGHKRZBFGoAg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"570a11be8de334fe7a92a4924dbb502d9b2086628711b4b09f8d6d9c7b9f1f41","last_reissued_at":"2026-07-05T02:05:32.414845Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T02:05:32.414845Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"BN-invariant sharpness regularizes the training model to better generalization","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Huishuai Zhang, Mingyang Yi, Tie-Yan Liu, Wei Chen, Zhi-Ming Ma","submitted_at":"2021-01-08T10:23:24Z","abstract_excerpt":"It is arguably believed that flatter minima can generalize better. However, it has been pointed out that the usual definitions of sharpness, which consider either the maxima or the integral of loss over a $\\delta$ ball of parameters around minima, cannot give consistent measurement for scale invariant neural networks, e.g., networks with batch normalization layer. In this paper, we first propose a measure of sharpness, BN-Sharpness, which gives consistent value for equivalent networks under BN. It achieves the property of scale invariance by connecting the integral diameter with the scale of p"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2101.02944","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/2101.02944/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":"2101.02944","created_at":"2026-07-05T02:05:32.414902+00:00"},{"alias_kind":"arxiv_version","alias_value":"2101.02944v1","created_at":"2026-07-05T02:05:32.414902+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2101.02944","created_at":"2026-07-05T02:05:32.414902+00:00"},{"alias_kind":"pith_short_12","alias_value":"K4FBDPUN4M2P","created_at":"2026-07-05T02:05:32.414902+00:00"},{"alias_kind":"pith_short_16","alias_value":"K4FBDPUN4M2P46US","created_at":"2026-07-05T02:05:32.414902+00:00"},{"alias_kind":"pith_short_8","alias_value":"K4FBDPUN","created_at":"2026-07-05T02:05:32.414902+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2604.00029","citing_title":"Spatio-Temporal Uncertainty-Modulated Physics-Informed Neural Networks for Solving Hyperbolic Conservation Laws with Strong Shocks","ref_index":14,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/K4FBDPUN4M2P46USUSJE3O2QFW","json":"https://pith.science/pith/K4FBDPUN4M2P46USUSJE3O2QFW.json","graph_json":"https://pith.science/api/pith-number/K4FBDPUN4M2P46USUSJE3O2QFW/graph.json","events_json":"https://pith.science/api/pith-number/K4FBDPUN4M2P46USUSJE3O2QFW/events.json","paper":"https://pith.science/paper/K4FBDPUN"},"agent_actions":{"view_html":"https://pith.science/pith/K4FBDPUN4M2P46USUSJE3O2QFW","download_json":"https://pith.science/pith/K4FBDPUN4M2P46USUSJE3O2QFW.json","view_paper":"https://pith.science/paper/K4FBDPUN","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2101.02944&json=true","fetch_graph":"https://pith.science/api/pith-number/K4FBDPUN4M2P46USUSJE3O2QFW/graph.json","fetch_events":"https://pith.science/api/pith-number/K4FBDPUN4M2P46USUSJE3O2QFW/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/K4FBDPUN4M2P46USUSJE3O2QFW/action/timestamp_anchor","attest_storage":"https://pith.science/pith/K4FBDPUN4M2P46USUSJE3O2QFW/action/storage_attestation","attest_author":"https://pith.science/pith/K4FBDPUN4M2P46USUSJE3O2QFW/action/author_attestation","sign_citation":"https://pith.science/pith/K4FBDPUN4M2P46USUSJE3O2QFW/action/citation_signature","submit_replication":"https://pith.science/pith/K4FBDPUN4M2P46USUSJE3O2QFW/action/replication_record"}},"created_at":"2026-07-05T02:05:32.414902+00:00","updated_at":"2026-07-05T02:05:32.414902+00:00"}