{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2022:WKFE55BDXPK56CH5J7YZVVJBQY","short_pith_number":"pith:WKFE55BD","schema_version":"1.0","canonical_sha256":"b28a4ef423bbd5df08fd4ff19ad521862b0589b4e43d30f2726d857e5c902236","source":{"kind":"arxiv","id":"2207.01156","version":1},"attestation_state":"computed","paper":{"title":"Removing Batch Normalization Boosts Adversarial Training","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.CV"],"primary_cat":"cs.LG","authors_text":"Aston Zhang, Haotao Wang, Mu Li, Shuai Zheng, Xingjian Shi, Zhangyang Wang","submitted_at":"2022-07-04T01:39:37Z","abstract_excerpt":"Adversarial training (AT) defends deep neural networks against adversarial attacks. One challenge that limits its practical application is the performance degradation on clean samples. A major bottleneck identified by previous works is the widely used batch normalization (BN), which struggles to model the different statistics of clean and adversarial training samples in AT. Although the dominant approach is to extend BN to capture this mixture of distribution, we propose to completely eliminate this bottleneck by removing all BN layers in AT. Our normalizer-free robust training (NoFrost) metho"},"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":"2207.01156","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2022-07-04T01:39:37Z","cross_cats_sorted":["cs.CV"],"title_canon_sha256":"b50762d6e05615ad884be8c325354266b63f146312f6c366bf9b15ab5be3e628","abstract_canon_sha256":"1cbd039acfc15b4e064a66dde9eac842224393d3fcf3057edd0cb3357b4ffe66"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T04:37:08.539493Z","signature_b64":"cTnVYHqj5R0JTGx/HSZfq0q28V1eE1cCBFgQfSYsNOlHdVOeAsrC3B4vNA4Yru3bbqc+G8ihwp7ynSrh+EawDQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"b28a4ef423bbd5df08fd4ff19ad521862b0589b4e43d30f2726d857e5c902236","last_reissued_at":"2026-07-05T04:37:08.539077Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T04:37:08.539077Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Removing Batch Normalization Boosts Adversarial Training","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.CV"],"primary_cat":"cs.LG","authors_text":"Aston Zhang, Haotao Wang, Mu Li, Shuai Zheng, Xingjian Shi, Zhangyang Wang","submitted_at":"2022-07-04T01:39:37Z","abstract_excerpt":"Adversarial training (AT) defends deep neural networks against adversarial attacks. One challenge that limits its practical application is the performance degradation on clean samples. A major bottleneck identified by previous works is the widely used batch normalization (BN), which struggles to model the different statistics of clean and adversarial training samples in AT. Although the dominant approach is to extend BN to capture this mixture of distribution, we propose to completely eliminate this bottleneck by removing all BN layers in AT. Our normalizer-free robust training (NoFrost) metho"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2207.01156","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/2207.01156/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":"2207.01156","created_at":"2026-07-05T04:37:08.539144+00:00"},{"alias_kind":"arxiv_version","alias_value":"2207.01156v1","created_at":"2026-07-05T04:37:08.539144+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2207.01156","created_at":"2026-07-05T04:37:08.539144+00:00"},{"alias_kind":"pith_short_12","alias_value":"WKFE55BDXPK5","created_at":"2026-07-05T04:37:08.539144+00:00"},{"alias_kind":"pith_short_16","alias_value":"WKFE55BDXPK56CH5","created_at":"2026-07-05T04:37:08.539144+00:00"},{"alias_kind":"pith_short_8","alias_value":"WKFE55BD","created_at":"2026-07-05T04:37:08.539144+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/WKFE55BDXPK56CH5J7YZVVJBQY","json":"https://pith.science/pith/WKFE55BDXPK56CH5J7YZVVJBQY.json","graph_json":"https://pith.science/api/pith-number/WKFE55BDXPK56CH5J7YZVVJBQY/graph.json","events_json":"https://pith.science/api/pith-number/WKFE55BDXPK56CH5J7YZVVJBQY/events.json","paper":"https://pith.science/paper/WKFE55BD"},"agent_actions":{"view_html":"https://pith.science/pith/WKFE55BDXPK56CH5J7YZVVJBQY","download_json":"https://pith.science/pith/WKFE55BDXPK56CH5J7YZVVJBQY.json","view_paper":"https://pith.science/paper/WKFE55BD","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2207.01156&json=true","fetch_graph":"https://pith.science/api/pith-number/WKFE55BDXPK56CH5J7YZVVJBQY/graph.json","fetch_events":"https://pith.science/api/pith-number/WKFE55BDXPK56CH5J7YZVVJBQY/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/WKFE55BDXPK56CH5J7YZVVJBQY/action/timestamp_anchor","attest_storage":"https://pith.science/pith/WKFE55BDXPK56CH5J7YZVVJBQY/action/storage_attestation","attest_author":"https://pith.science/pith/WKFE55BDXPK56CH5J7YZVVJBQY/action/author_attestation","sign_citation":"https://pith.science/pith/WKFE55BDXPK56CH5J7YZVVJBQY/action/citation_signature","submit_replication":"https://pith.science/pith/WKFE55BDXPK56CH5J7YZVVJBQY/action/replication_record"}},"created_at":"2026-07-05T04:37:08.539144+00:00","updated_at":"2026-07-05T04:37:08.539144+00:00"}