{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:CK45O3AIWOXDAKQRXL7NTTX546","short_pith_number":"pith:CK45O3AI","schema_version":"1.0","canonical_sha256":"12b9d76c08b3ae302a11bafed9cefde7bd029c772952c67922b6efce7118e366","source":{"kind":"arxiv","id":"2505.11370","version":3},"attestation_state":"computed","paper":{"title":"Understanding Nonlinear Implicit Bias via Region Counts in Input Space","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Huishuai Zhang, Jingwei Li, Jing Xu, Jingzhao Zhang, Zifan Wang","submitted_at":"2025-05-16T15:34:19Z","abstract_excerpt":"One explanation for the strong generalization ability of neural networks is implicit bias. Yet, the definition and mechanism of implicit bias in non-linear contexts remains little understood. In this work, we propose to characterize implicit bias by the count of connected regions in the input space with the same predicted label. Compared with parameter-dependent metrics (e.g., norm or normalized margin), region count can be better adapted to nonlinear, overparameterized models, because it is determined by the function mapping and is invariant to reparametrization. Empirically, we found that sm"},"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":"2505.11370","kind":"arxiv","version":3},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2025-05-16T15:34:19Z","cross_cats_sorted":[],"title_canon_sha256":"e30bfedd0ece9d2a9f6fc213e0b88fc7aeea85847e0e10ce943df8735dc82df3","abstract_canon_sha256":"ce0f92c14eb546f963c8d0e18e303baec49d8239f89f2864ba6875c12ae3c9f1"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:53:02.947554Z","signature_b64":"IUlBdmgCAMK60jr/bFP02wIYSavQI0qXl+Mv8k8A5RuErK/bGdV6ktvr4Wvg2k/VKSl4YE7kx/m0JvdBhQouAw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"12b9d76c08b3ae302a11bafed9cefde7bd029c772952c67922b6efce7118e366","last_reissued_at":"2026-07-05T11:53:02.947096Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:53:02.947096Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Understanding Nonlinear Implicit Bias via Region Counts in Input Space","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Huishuai Zhang, Jingwei Li, Jing Xu, Jingzhao Zhang, Zifan Wang","submitted_at":"2025-05-16T15:34:19Z","abstract_excerpt":"One explanation for the strong generalization ability of neural networks is implicit bias. Yet, the definition and mechanism of implicit bias in non-linear contexts remains little understood. In this work, we propose to characterize implicit bias by the count of connected regions in the input space with the same predicted label. Compared with parameter-dependent metrics (e.g., norm or normalized margin), region count can be better adapted to nonlinear, overparameterized models, because it is determined by the function mapping and is invariant to reparametrization. Empirically, we found that sm"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2505.11370","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/2505.11370/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":"2505.11370","created_at":"2026-07-05T11:53:02.947154+00:00"},{"alias_kind":"arxiv_version","alias_value":"2505.11370v3","created_at":"2026-07-05T11:53:02.947154+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2505.11370","created_at":"2026-07-05T11:53:02.947154+00:00"},{"alias_kind":"pith_short_12","alias_value":"CK45O3AIWOXD","created_at":"2026-07-05T11:53:02.947154+00:00"},{"alias_kind":"pith_short_16","alias_value":"CK45O3AIWOXDAKQR","created_at":"2026-07-05T11:53:02.947154+00:00"},{"alias_kind":"pith_short_8","alias_value":"CK45O3AI","created_at":"2026-07-05T11:53:02.947154+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/CK45O3AIWOXDAKQRXL7NTTX546","json":"https://pith.science/pith/CK45O3AIWOXDAKQRXL7NTTX546.json","graph_json":"https://pith.science/api/pith-number/CK45O3AIWOXDAKQRXL7NTTX546/graph.json","events_json":"https://pith.science/api/pith-number/CK45O3AIWOXDAKQRXL7NTTX546/events.json","paper":"https://pith.science/paper/CK45O3AI"},"agent_actions":{"view_html":"https://pith.science/pith/CK45O3AIWOXDAKQRXL7NTTX546","download_json":"https://pith.science/pith/CK45O3AIWOXDAKQRXL7NTTX546.json","view_paper":"https://pith.science/paper/CK45O3AI","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2505.11370&json=true","fetch_graph":"https://pith.science/api/pith-number/CK45O3AIWOXDAKQRXL7NTTX546/graph.json","fetch_events":"https://pith.science/api/pith-number/CK45O3AIWOXDAKQRXL7NTTX546/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/CK45O3AIWOXDAKQRXL7NTTX546/action/timestamp_anchor","attest_storage":"https://pith.science/pith/CK45O3AIWOXDAKQRXL7NTTX546/action/storage_attestation","attest_author":"https://pith.science/pith/CK45O3AIWOXDAKQRXL7NTTX546/action/author_attestation","sign_citation":"https://pith.science/pith/CK45O3AIWOXDAKQRXL7NTTX546/action/citation_signature","submit_replication":"https://pith.science/pith/CK45O3AIWOXDAKQRXL7NTTX546/action/replication_record"}},"created_at":"2026-07-05T11:53:02.947154+00:00","updated_at":"2026-07-05T11:53:02.947154+00:00"}