{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2019:YD4L4FAKOVC2XMFZJYAA6SR5ZJ","short_pith_number":"pith:YD4L4FAK","schema_version":"1.0","canonical_sha256":"c0f8be140a7545abb0b94e000f4a3dca6aff86a7360d5274efa4ac55c230fa17","source":{"kind":"arxiv","id":"1904.12843","version":2},"attestation_state":"computed","paper":{"title":"Adversarial Training for Free!","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.CR","cs.CV","stat.ML"],"primary_cat":"cs.LG","authors_text":"Ali Shafahi, Amin Ghiasi, Christoph Studer, Gavin Taylor, John Dickerson, Larry S. Davis, Mahyar Najibi, Tom Goldstein, Zheng Xu","submitted_at":"2019-04-29T17:50:32Z","abstract_excerpt":"Adversarial training, in which a network is trained on adversarial examples, is one of the few defenses against adversarial attacks that withstands strong attacks. Unfortunately, the high cost of generating strong adversarial examples makes standard adversarial training impractical on large-scale problems like ImageNet. We present an algorithm that eliminates the overhead cost of generating adversarial examples by recycling the gradient information computed when updating model parameters. Our \"free\" adversarial training algorithm achieves comparable robustness to PGD adversarial training on 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":"1904.12843","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2019-04-29T17:50:32Z","cross_cats_sorted":["cs.CR","cs.CV","stat.ML"],"title_canon_sha256":"033d00e61105aab17dec84e4a047ccd4f585e51ac127cf136230527569668d1c","abstract_canon_sha256":"835acd889d7764a9b15262fee6e931c887ba97813bda94ec3e01d4255baf0b11"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T00:20:52.360125Z","signature_b64":"O26DykwTtuLXr3LWiLT0M0iB/MK8BImUV/kKMy2lIUnp9W7MsV+oy8o0NbzT/jFP64PzYzWUVKhT3NmdeLCvCw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"c0f8be140a7545abb0b94e000f4a3dca6aff86a7360d5274efa4ac55c230fa17","last_reissued_at":"2026-07-05T00:20:52.359739Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T00:20:52.359739Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Adversarial Training for Free!","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.CR","cs.CV","stat.ML"],"primary_cat":"cs.LG","authors_text":"Ali Shafahi, Amin Ghiasi, Christoph Studer, Gavin Taylor, John Dickerson, Larry S. Davis, Mahyar Najibi, Tom Goldstein, Zheng Xu","submitted_at":"2019-04-29T17:50:32Z","abstract_excerpt":"Adversarial training, in which a network is trained on adversarial examples, is one of the few defenses against adversarial attacks that withstands strong attacks. Unfortunately, the high cost of generating strong adversarial examples makes standard adversarial training impractical on large-scale problems like ImageNet. We present an algorithm that eliminates the overhead cost of generating adversarial examples by recycling the gradient information computed when updating model parameters. Our \"free\" adversarial training algorithm achieves comparable robustness to PGD adversarial training on th"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"1904.12843","kind":"arxiv","version":2},"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/1904.12843/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":"1904.12843","created_at":"2026-07-05T00:20:52.359799+00:00"},{"alias_kind":"arxiv_version","alias_value":"1904.12843v2","created_at":"2026-07-05T00:20:52.359799+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.1904.12843","created_at":"2026-07-05T00:20:52.359799+00:00"},{"alias_kind":"pith_short_12","alias_value":"YD4L4FAKOVC2","created_at":"2026-07-05T00:20:52.359799+00:00"},{"alias_kind":"pith_short_16","alias_value":"YD4L4FAKOVC2XMFZ","created_at":"2026-07-05T00:20:52.359799+00:00"},{"alias_kind":"pith_short_8","alias_value":"YD4L4FAK","created_at":"2026-07-05T00:20:52.359799+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":3,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2607.00207","citing_title":"Homogenization of $\\ell_2$-Adversarial Training in High-Dimensions: Exact Dynamics under Stochastic Gradient Descent","ref_index":61,"is_internal_anchor":false},{"citing_arxiv_id":"2605.14868","citing_title":"Fast Adversarial Attacks with Gradient Prediction","ref_index":9,"is_internal_anchor":false},{"citing_arxiv_id":"2606.00738","citing_title":"SORA: Free Second-Order Attacks in Fast Adversarial Training","ref_index":10,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/YD4L4FAKOVC2XMFZJYAA6SR5ZJ","json":"https://pith.science/pith/YD4L4FAKOVC2XMFZJYAA6SR5ZJ.json","graph_json":"https://pith.science/api/pith-number/YD4L4FAKOVC2XMFZJYAA6SR5ZJ/graph.json","events_json":"https://pith.science/api/pith-number/YD4L4FAKOVC2XMFZJYAA6SR5ZJ/events.json","paper":"https://pith.science/paper/YD4L4FAK"},"agent_actions":{"view_html":"https://pith.science/pith/YD4L4FAKOVC2XMFZJYAA6SR5ZJ","download_json":"https://pith.science/pith/YD4L4FAKOVC2XMFZJYAA6SR5ZJ.json","view_paper":"https://pith.science/paper/YD4L4FAK","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=1904.12843&json=true","fetch_graph":"https://pith.science/api/pith-number/YD4L4FAKOVC2XMFZJYAA6SR5ZJ/graph.json","fetch_events":"https://pith.science/api/pith-number/YD4L4FAKOVC2XMFZJYAA6SR5ZJ/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/YD4L4FAKOVC2XMFZJYAA6SR5ZJ/action/timestamp_anchor","attest_storage":"https://pith.science/pith/YD4L4FAKOVC2XMFZJYAA6SR5ZJ/action/storage_attestation","attest_author":"https://pith.science/pith/YD4L4FAKOVC2XMFZJYAA6SR5ZJ/action/author_attestation","sign_citation":"https://pith.science/pith/YD4L4FAKOVC2XMFZJYAA6SR5ZJ/action/citation_signature","submit_replication":"https://pith.science/pith/YD4L4FAKOVC2XMFZJYAA6SR5ZJ/action/replication_record"}},"created_at":"2026-07-05T00:20:52.359799+00:00","updated_at":"2026-07-05T00:20:52.359799+00:00"}